Publications
2022
Conference Papers
Dimitrios Kavelidis Frantzis, Emmanouil Tsardoulias, Thomas Karanikiotis, Eleni Poptsi, Magda Tsolaki and Andreas Symeonidis
National Conference ACOUSTICS 2022, 2022 Oct
  In this study, the validity of a Machine Learning multiclass classification process is examined, as to classify a speaker in a cognitive decline stage, aiming to develop a simple screening test. The target classes comprise Cognitively Healthy controls, Subjective Cognitive Decline and Early & Late Mild Cognitive Impairment. Speech data was collected from structured interviews on 84 people, split in stages of increasing required levels of cognitive difficulty. Audio features were extracted based on Silence, Prosody and Zero-Crossings, as well as on the feature vectors’ differences between stages, and were evaluated with the Random Forest, Extra-Trees and Support Vector Machines classifiers. The best classification was achieved using models trained with stage differences features (on SVM), resulting in a mean accuracy of 80.99±3.29%. @conference{2022kavAlzSpeech, | |
Eleni Poptsi, Despoina Moraitou, Emmanouil Tsardoulias, Andreas Symeonidis and Magda Tsolaki
"Υποκειμενική νοητική εξασθένιση: Κομμάτι της υγιούς γήρανσης ή έναρξη νευροεκφύλισης; Νεότερα δεδομένα της συστοιχίας R4Alz"
8ο Παγκρήτιο Διεπιστημονικό Συνέδριο Νόσου Alzheimer και Συναφών Διαταραχών και 4ο Πανελλήνιο Συνέδριο στην ενεργό και υγιή γήρανση, Σεπτεμβρίος 2022, Εμπορικό και Βιομηχανικό Επιμελητήριο Ηρακλείου, 2022 Sep
@conference{2022Kretepub1, | |
Evangelos Papathomas, Themistoklis Diamantopoulos and Andreas Symeonidis
"Semantic Code Search in Software Repositories using Neural Machine Translation"
Fundamental Approaches to Software Engineering, pp. 225-244, Springer International Publishing, Cham, 2022 Apr
 Nowadays, software development is accelerated through the reuse of code snippets found online in question-answering platforms and software repositories. In order to be efficient, this process requires forming an appropriate query and identifying the most suitable code snippet, which can sometimes be challenging and particularly time-consuming. Over the last years, several code recommendation systems have been developed to offer a solution to this problem. Nevertheless, most of them recommend API calls or sequences instead of reusable code snippets. Furthermore, they do not employ architectures advanced enough to exploit the semantics of natural language and code in order to form the optimal query from the question posed. To overcome these issues, we propose CodeTransformer, a code recommendation system that provides useful, reusable code snippets extracted from open-source GitHub repositories. By employing a neural network architecture that comprises advanced attention mechanisms, our system effectively understands and models natural language queries and code snippets in a joint vector space. Upon evaluating CodeTransformer quantitatively against a similar system and qualitatively using a dataset from Stack Overflow, we conclude that our approach can recommend useful and reusable snippets to developers. @conference{FASE2022, | |
Andreas Goulas, Nikolaos Malamas and Andreas L. Symeonidis
"A Methodology for Enabling NLP Capabilities on Edge and Low-Resource Devices"
Natural Language Processing and Information Systems, pp. 197--208, Springer International Publishing, Cham, 2022 Jun
Conversational assistants with increasing NLP capabilities are becoming commodity functionality for most new devices. However, the underlying language models responsible for language-related intelligence are typically characterized by a large number of parameters and high demand for memory and resources. This makes them a no-go for edge and low-resource devices, forcing them to be cloud-hosted, hence experiencing delays. To this end, we design a systematic language-agnostic methodology to develop powerful lightweight NLP models using knowledge distillation techniques, this way building models suitable for such low resource devices. We follow the steps of the proposed approach for the Greek language and build the first - to the best of our knowledge - lightweight Greek language model, which we make publicly available. We train and evaluate GloVe word embeddings in Greek and efficiently distill Greek-BERT into various BiLSTM models, without considerable loss in performance. Experiments indicate that knowledge distillation and data augmentation can improve the performance of simple BiLSTM models for two NLP tasks in Modern Greek, i.e., Topic Classification and Natural Language Inference, making them suitable candidates for low-resource devices. @inproceedings{goulas-et-al, | |
Argyrios Papoudakis, Thomas Karanikiotis and Andreas Symeonidis
"A Mechanism for Automatically Extracting Reusable and Maintainable Code Idioms from Software Repositories"
Proceedings of the 17th International Conference on Software Technologies - ICSOFT, pp. 79-90, SciTePress, 2022 Jul
The importance of correct, qualitative and evolvable code is non-negotiable, when considering the maintainability potential of software. At the same time, the deluge of software residing in code hosting platforms has led to a new component-based software development paradigm, where reuse of suitable software components/ snippets is important for software projects to be implemented as fast as possible. However, ensuring acceptable quality that will guarantee basic maintainability is also required. A condition for acceptable software reusability and maintainability is the use of idiomatic code, based on syntactic fragments that recur frequently across software projects and are characterized by high quality. In this work, we present a mechanism that employs the top repositories from GitHub in order to automatically identify reusable and maintainable code idioms. By extracting the Abstract Syntax Tree representation of each project we group code snippets that appear to have similar struc tural and semantic information. Preliminary evaluation of our methodology indicates that our approach can identify commonly used, reusable and maintainable code idioms that can be effectively given as actionable recommendations to the developers. @conference{icsoft22karanikiotis, | |
Georgios Kalantzis, Gerasimos Papakostas, Thomas Karanikiotis, Michail Papamichail and Andreas Symeonidis
"A Heuristic Approach towards Continuous Implicit Authentication"
2022 IEEE International Joint Conference on Biometrics (IJCB), pp. 1-7, IEEE, 2022 Oct
Smartphones nowadays handle large amounts of sensitive user information, since users exchange undisclosed information on an everyday basis. This generates the need for more effective authentication mechanisms, deviating from the traditional ones. In this direction, many research approaches are targeted towards continuous implicit authentication, on the basis of modelling the constant interaction of the user with the device. These approaches yield promising results, however certain improvements can be made by exploiting the sequential order of the predictions and the known performance metrics. In this work, we propose a heuristics algorithm, which, given a series of predictions from any continuous implicit authentication model, can ex-ploit the sequential order in order to fix any false predictions and improve the accuracy of the smartphone security system. Preliminary evaluation on several axes indicates that our approach can effectively improve any CIA model and achieve significantly better results. @conference{ijcb2022karanikiotis, | |
Eleni Poptsi, Despoina Moraitou, Emmanouil Tsardoulias, Andreas Symeonidis and Magda Tsolaki
"Νευροψυχολογική συστοιχία REMEDES for Alzheimer (R4Alz): Παρουσίαση ενός εργαλείου πρώιμης διάγνωσης των νευροεκφυλιστικών νοσημάτων"
8ο Παγκρήτιο Διεπιστημονικό Συνέδριο Νόσου Alzheimer και Συναφών Διαταραχών και 4ο Πανελλήνιο Συνέδριο στην ενεργό και υγιή γήρανση, Σεπτεμβρίος 2022, Εμπορικό και Βιομηχανικό Επιμελητήριο Ηρακλείου, 2022 Sep
@conference{Kreteconf2_2022, | |
Emmanouil Tsardoulias, Eleni Poptsi, Dimitrios F. Kavelidis, Thomas Karanikiotis, Magda Tsolaki, Despoina Moraitou and Andreas L. Symeonidis
"Early detection of neurocognitive decline using Cyber Physical Systems and Artificial Intelligence"
9th Technology Forum, Thessaloniki, 2022 Sep
@conference{tf20221, | |
Theodoros Papafotiou, Efthymia Amarantidou, Efseveia Nestoropoulou and Emmanouil Tsardoulias
"Autonomous Driving Vehicle in 1:10 scaled environment"
9th Technology Forum, Thessaloniki, 2022 Sep
@conference{tf20222, | |
Konstantinos Panayiotou, Emmanouil Tsardoulias and Andreas Symeonidis
"Low-code development & verification of Cyber-Physical Systems"
9th Technology Forum, Thessaloniki, 2022 Sep
@conference{tf20223, |
2021
Conference Papers
Eleni Poptsi, Despina Moraitou, Emmanouil Tsardoulias, Andreas L. Symeonidis and Magda Tsolaki
"Είναι εφικτός ο διαχωρισμός του Υγιούς Νοητικά Γήρατος από την Υποκειμενική Νοητική Εξασθένιση; Πιλοτικά αποτελέσματα της καινοτόμου συστοιχίας REMEDES for Alzheimer (R4Alz)"
12th Panhellenic Conference on Alzheimer's Disease & 3rdMediterranean Conference on Neurodegenerative Diseases PICAD & MeCoND, Thessaloniki,Greece, 2021 Feb
@conference{elena2021alzconf, | |
E. Tsardoulias, C. Zolotas, S. Siouli, P. Antoniou, S. Amanatiadis, T. Karanikiotis, E. Chondromatidis, P. Bamidis, G. Karagiannis and A. Symeonidis
"Science and mathematics education via remote robotics deployment - The TekTrain paradigm"
14th annual International Conference of Education, Research and Innovation - 2021, 2021 Nov
@conference{ets2021iceriTektrain, | |
Themistoklis Diamantopoulos, Christiana Galegalidou and Andreas L. Symeonidis
"Software Task Importance Prediction based on Project Management Data"
Proceedings of the 16th International Conference on Software Technologies (ICSOFT 2021), pp. 269-276, 2021 Jul
With the help of project management tools and code hosting facilities, software development has been transformed into an easy-to-decentralize business. However, determining the importance of tasks within a software engineering process in order to better prioritize and act on has always been an interesting challenge. Although several approaches on bug severity/priority prediction exist, the challenge of task importance prediction has not been sufficiently addressed in current research. Most approaches do not consider the meta-data and the temporal characteristics of the data, while they also do not take into account the ordinal characteristics of the importance/severity variable. In this work, we analyze the challenge of task importance prediction and propose a prototype methodology that extracts both textual (titles, descriptions) and meta-data (type, assignee) characteristics from tasks and employs a sliding window technique to model their time frame. After that, we evaluate three different prediction methods, a multi-class classifier, a regression algorithm, and an ordinal classification technique, in order to assess which model is the most effective for encompassing the relative ordering between different importance values. The results of our evaluation are promising, leaving room for future research. @conference{ICSOFT2021, | |
Thomas Karanikiotis, Kyriakos C. Chatzidimitriou and Andreas L. Symeonidis
"Towards Automatically Generating a Personalized Code Formatting Mechanism"
Proceedings of the 16th International Conference on Software Technologies - ICSOFT, pp. 90-101, SciTePress, 2021 Jul
Source code readability and comprehensibility are continuously gaining interest, due to the wide adoption of component-based software development and the (re)use of software residing in code hosting platforms. Consistent code styling and formatting across a project tend to improve readability, while most code formatting approaches rely on a set of rules defined by experts, that aspire to model a commonly accepted formatting. This approach is usually based on the experts’ expertise, is time-consuming and does not take into account the way a team develops software. Thus, it becomes too intrusive and, in many cases, is not adopted. In this work we present an automated mechanism, that, given a set of source code files, can be trained to recognize the formatting style used across a project and identify deviations, in a completely unsupervised manner. At first, source code is transformed into small meaningful pieces, called tokens, which are used to train the models of our mechanism, in or der to predict the probability of a token being wrongly positioned. Preliminary evaluation on various axes indicates that our approach can effectively detect formatting deviations from the project’s code styling and provide actionable recommendations to the developer. @conference{icsoft2021Codrep, | |
Antonis Dimitriou, Anastasios Tzitzis, Alexandros Filotheou, Spyros Megalou, Stavroula Siachalou, Aristidis R. Chatzistefanou, Andreana Malama, Emmanouil Tsardoulias, Konstantinos Panayiotou, Evaggelos Giannelos, Thodoris Vasiliadis, Ioannis Mouroutsos, Ioannis Karanikas, Loukas Petrou, Andreas Symeonidis, John Sahalos, Traianos Yioultsis and Aggelos Bletsas
"Autonomous Robots, Drones and Repeaters for Fast, Reliable, Low-Cost RFID Inventorying and Localization"
2021 6th International Conference on Smart and Sustainable Technologies (SpliTech), 2021 Sep
@conference{tsa2021rfidSplitech, |
2020
Conference Papers
Nikolaos L. Tsakiridis, Themistoklis Diamantopoulos, Andreas L. Symeonidis, John B. Theocharis, Athanasios Iossifides, Periklis Chatzimisios, George Pratos and Dimitris Kouvas
"Versatile Internet of Things for Agriculture: An eXplainable AI Approach"
International Conference on Artificial Intelligence Applications and Innovations, 2020 Jun
The increase of the adoption of IoT devices and the contemporary problem of food production have given rise to numerous applications of IoT in agriculture. These applications typically comprise a set of sensors that are installed in open fields and measure metrics, such as temperature or humidity, which are used for irrigation control systems. Though useful, most contemporary systems have high installation and maintenance costs, and they do not offer automated control or, if they do, they are usually not interpretable, and thus cannot be trusted for such critical applications. In this work, we design Vital, a system that incorporates a set of low-cost sensors, a robust data store, and most importantly an explainable AI decision support system. Our system outputs a fuzzy rule-base, which is interpretable and allows fully automating the irrigation of the fields. Upon evaluating Vital in two pilot cases, we conclude that it can be effective for monitoring open-field installations. @conference{AIAI2020, | |
Thomas Karanikiotis, Michail D. Papamichail, Kyriakos C. Chatzidimitriou, Napoleon-Christos I. Oikonomou, Andreas L. Symeonidis, and Sashi K. Saripalle
"Continuous Implicit Authentication through Touch Traces Modelling"
20th International Conference on Software Quality, Reliability and Security (QRS), pp. 111-120, 2020 Nov
Nowadays, the continuously increasing use of smartphones as the primary way of dealing with day-to-day tasks raises several concerns mainly focusing on privacy and security. In this context and given the known limitations and deficiencies of traditional authentication mechanisms, a lot of research efforts are targeted towards continuous implicit authentication on the basis of behavioral biometrics. In this work, we propose a methodology towards continuous implicit authentication that refrains from the limitations imposed by small-scale and/or controlled environment experiments by employing a real-world application used widely by a large number of individuals. Upon constructing our models using Support Vector Machines, we introduce a confidence-based methodology, in order to strengthen the effectiveness and the efficiency of our approach. The evaluation of our methodology on a set of diverse scenarios indicates that our approach achieves good results both in terms of efficiency and usability. @inproceedings{ciaQRS2020, | |
Themistoklis Diamantopoulos, Nikolaos Oikonomou and Andreas Symeonidis
"Extracting Semantics from Question-Answering Services for Snippet Reuse"
Fundamental Approaches to Software Engineering, pp. 119-139, Springer International Publishing, Cham, 2020 Apr
Nowadays, software developers typically search online for reusable solutions to common programming problems. However, forming the question appropriately, and locating and integrating the best solution back to the code can be tricky and time consuming. As a result, several mining systems have been proposed to aid developers in the task of locating reusable snippets and integrating them into their source code. Most of these systems, however, do not model the semantics of the snippets in the context of source code provided. In this work, we propose a snippet mining system, named StackSearch, that extracts semantic information from Stack Overlow posts and recommends useful and in-context snippets to the developer. Using a hybrid language model that combines Tf-Idf and fastText, our system effectively understands the meaning of the given query and retrieves semantically similar posts. Moreover, the results are accompanied with useful metadata using a named entity recognition technique. Upon evaluating our system in a set of common programming queries, in a dataset based on post links, and against a similar tool, we argue that our approach can be useful for recommending ready-to-use snippets to the developer. @conference{FASE2020, | |
Thomas Karanikiotis, Michail D. Papamichail, Giannis Gonidelis, Dimitra Karatza and Andreas L. Symeonidis
"A Data-driven Methodology towards Interpreting Readability against Software Properties"
Proceedings of the 15th International Conference on Software Technologies - ICSOFT, pp. 61-72, SciTePress, 2020 Jan
In the context of collaborative, agile software development, where effective and efficient software maintenance is of utmost importance, the need to produce readable source code is evident. Towards this direction, several approaches aspire to assess the extent to which a software component is readable. Most of them rely on experts who are responsible for determining the ground truth and/or set custom evaluation criteria, leading to results that are context-dependent and subjective. In this work, we employ a large set of static analysis metrics along with various coding violations towards interpreting readability as perceived by developers. In an effort to provide a fully automated and extendible methodology, we refrain from using experts; rather we harness data residing in online code hosting facilities towards constructing a dataset that includes more than one million methods that cover diverse development scenarios. After performing clustering based on source code size, we employ S upport Vector Regression in order to interpret the extent to which a software component is readable on three axes: complexity, coupling, and documentation. Preliminary evaluation on several axes indicates that our approach effectively interprets readability as perceived by developers against the aforementioned three primary source code properties. @inproceedings{karanikiotisICSOFT2020, | |
Themistoklis Diamantopoulos, Michail D. Papamichail, Thomas Karanikiotis, Kyriakos C. Chatzidimitriou and Andreas L. Symeonidis
"Employing Contribution and Quality Metrics for Quantifying the Software Development Process"
The 17th International Conference on Mining Software Repositories (MSR 2020), 2020 Jun
The full integration of online repositories in the contemporary software development process promotes remote work and remote collaboration. Apart from the apparent benefits, online repositories offer a deluge of data that can be utilized to monitor and improve the software development process. Towards this direction, we have designed and implemented a platform that analyzes data from GitHub in order to compute a series of metrics that quantify the contributions of project collaborators, both from a development as well as an operations (communication) perspective. We analyze contributions in an evolutionary manner throughout the projects' lifecycle and track the number of coding violations generated, this way aspiring to identify cases of software development that need closer monitoring and (possibly) further actions to be taken. In this context, we have analyzed the 3000 most popular Java GitHub projects and provide the data to the community. @conference{MSR2020, | |
Eleni Poptsi, Despina Moraitou, Emmanouil Tsardoulias, Andreas L. Symeonidis and Magda Tsolaki
"Towards novel tools for discriminating healthy adults from people with neurocognitivedisorders: A pilot study utilizing the REMEDES for Alzheimer (R4Alz) battery"
2020 Alzheimer's Disease International Conference, 2020 Dec
Background: The early diagnosis of neurocognitive disorders before the onset of the symptoms of the clinical diagnosis is the ultimate goal of the scientific community. REMEDES for Alzheimer (R4Alz) is a battery, designed for assessing cognitive control abilities in people with minor and major neurocognitive disorders. The battery utilizes the “REMEDES” system, capable of measuring reflexes using visual and auditory triggers. The battery comprises three (3) tasks for assessing working memory capacity, attention control and inhibitory control, plus cognitive flexibility. Objectives: To investigate (a) whether the R4Alz battery’s tasks differentiate healthy adults controls (HAc) aged 20-85 years old from people with Subjective Cognitive Decline (SCD) and Mild Cognitive Impairment (MCI), (b) whether the battery is free of age, gender and educational level effects, and (c) the criterion-related validity of the R4Alz in all groups. Methods: The R4Alz battery administered in 100 Greek adults, categorized in healthy adult controls (HAc) (n = 39), community-dwelling older adults with SCD (n = 25) and patients with MCI (n = 36). Statistical analysis comprised Analysis of Variance (ANOVA) and Multivariate Analysis of Covariance (MANCOVA) with age and demographics as covariates when was necessary. The Scheffe post hoc test, applied in batteries’ tasks, as well. Pearson’s Correlation was also used for the investigation of the criterion-related validity. Results: The updating of working memory task discriminates the three groups and is free of gender (p = 0.184), age (p = 0.280) and education (p = 0.367) effects. The attention control task also discriminates the three diagnostic groups, while is independent from gender (p = 0.465) and education (p = 0.061). The inhibition control task is also gender (p = 0.697), age (p = 0.604) and education (p = 0.111) independent and can discriminate HAc from MCI and SCD from MCI. Criterion-related validity in all groups was supported by significant correlations. The updating of working memory task was correlated with the n-back test, where the attention control task was correlated with the Paper and pencil Dual test and the Test of Everyday Αttention (TEA). Finally the inhibition control task of the R4Alz battery was correlated with the Color-Word Interference Test of D-KEFS. Conclusion: The preliminary data of this study indicates that the R4Alz battery is a novel technological approach regarding the psychometric assessment of people with minor and major cognitive deficits, since it is free of demographic effects and it can help with discriminating HAc from SCI and MCI and SCI from MCI. @conference{poptsiadi2020, | |
Vasileios Matsoukas, Themistoklis Diamantopoulos, Michail D. Papamichail and Andreas L. Symeonidis
"Towards Analyzing Contributions from Software Repositories to Optimize Issue Assignment"
Proceedings of the 2020 IEEE International Conference on Software Quality, Reliability and Security (QRS), IEEE, Vilnius, Lithuania, 2020 Jul
Most software teams nowadays host their projects online and monitor software development in the form of issues/tasks. This process entails communicating through comments and reporting progress through commits and closing issues. In this context, assigning new issues, tasks or bugs to the most suitable contributor largely improves efficiency. Thus, several automated issue assignment approaches have been proposed, which however have major limitations. Most systems focus only on assigning bugs using textual data, are limited to projects explicitly using bug tracking systems, and may require manually tuning parameters per project. In this work, we build an automated issue assignment system for GitHub, taking into account the commits and issues of the repository under analysis. Our system aggregates feature probabilities using a neural network that adapts to each project, thus not requiring manual parameter tuning. Upon evaluating our methodology, we conclude that it can be efficient for automated issue assignment. @conference{QRS2020IssueAssignment, | |
Anastasios Tzitzis, Alexandros Filotheou, Stavroula Siachalou, Emmanouil Tsardoulias, Spyros Megalou, Aggelos Bletsas, Konstantinos Panayiotou, Andreas Symeonidis, Traianos Yioultsis and Antonis G. Dimitriou
"Real-time 3D localization of RFID-tagged products by ground robots and drones with commercial off-the-shelf RFID equipment: Challenges and Solutions"
2020 IEEE International Conference on RFID (RFID), 2020 Oct
In this paper we investigate the problem of localizing passive RFID tags by ground robots and drones. We focus on autonomous robots, capable of entering a previously unknown environment, creating a 3D map of it, navigating safely in it, localizing themselves while moving, then localizing all RFID tagged objects and pinpointing their locations in the 3D map with cm accuracy. To the best of our knowledge, this is the first paper that presents the complex joint problem, including challenges from the field of robotics - i) sensors utilization, ii) local and global path planners, iii) navigation, iv) simultaneous localization of the robot and mapping - and from the field of RFIDs - vi) localization of the tags. We restrict our analysis to solutions, involving commercial UHF EPC Gen2 RFID tags, commercial off-the-self RFID readers and 3D real-time-only methods for tag-localization. We briefly present a new method, suitable for real-time 3D inventorying, and compare it with our two recent methods. Comparison is carried out on a new set of experiments, conducted in a multipath-rich indoor environment, where the actual problem is treated; i.e. our prototype robot constructs a 3D map, navigates in the environment, continuously estimates its poses as well as the locations of the surrounding tags. Localization results are given in a few seconds for 100 tags, parsing approximately 100000 measured samples from 4 antennas, collected within 4 minutes and achieving a mean 3D error of 25cm, which includes the error propagating from robotics and the uncertainty related to the "ground truth" of the tags' placement. @conference{tzitzis2020realtime, |
2019
Conference Papers
Kyriakos C Chatzidimitriou, Michail D Papamichail, Napoleon-Christos I Oikonomou, Dimitrios Lampoudis and Andreas L Symeonidis
"Cenote: A Big Data Management and Analytics Infrastructure for the Web of Things"
IEEE/WIC/ACM International Conference on Web Intelligence, pp. 282-285, ACM, 2019 Oct
In the era of Big Data, Cloud Computing and Internet of Things, most of the existing, integrated solutions that attempt to solve their challenges are either proprietary, limit functionality to a predefined set of requirements, or hide the way data are stored and accessed. In this work we propose Cenote, an open source Big Data management and analytics infrastructure for the Web of Things that overcomes the above limitations. Cenote is built on component-based software engineering principles and provides an all-inclusive solution based on components that work well individually. @inproceedings{Chatzidimitriou:2019:CBD:3350546.3352531, | |
Themistoklis Diamantopoulos, Maria-Ioanna Sifaki and Andreas L. Symeonidis
"Towards Mining Answer Edits to Extract Evolution Patterns in Stack Overflow"
16th International Conference on Mining Software Repositories, 2019 Mar
Thecurrentstateofpracticedictatesthatinorderto solve a problem encountered when building software, developers ask for help in online platforms, such as Stack Overflow. In this context of collaboration, answers to question posts often undergo several edits to provide the best solution to the problem stated. In this work, we explore the potential of mining Stack Overflow answer edits to extract common patterns when answering a post. In particular, we design a similarity scheme that takes into account the text and code of answer edits and cluster edits according to their semantics. Upon applying our methodology, we provide frequent edit patterns and indicate how they could be used to answer future research questions. Our evaluation indicates that our approach can be effective for identifying commonly applied edits, thus illustrating the transformation path from the initial answer to the optimal solution. @conference{Diamantopoulos2019, | |
Tsardoulias Emmanouil, Panayiotou Konstantinos, Symeonidis Andreas and Petrou Loukas
"REMEDES: Τεχνικά χαρακτηριστικά και προδιαγραφές συστήματος αποτίμησης κιναισθησίας προς διάγνωση της νόσου Alzheimer"
11th Panhellenic Conference on Alzheimer's Disease & 3rd Mediterranean Conference on Neurodegenerative Diseases PICAD & MeCoND, Thessaloniki, Greece, 2019 Feb
Το REMEDES αποτελεί ένα σύστημα προσανατολισμένο στην μέτρηση και καταγραφή αντανακλαστικών και αντίδρασης με υψηλή ακρίβεια, κάνοντας χρήση οπτικών ή/και ακουστικών ερεθισμάτων. Το σύστημα είναι κατάλληλο για την ποσοτικοποίηση της ιδιοδεκτικότητας/κιναισθησίας, καθώς στηρίζεται στο βασικό πεδίο της ανθρώπινης δράσης/αντίδρασης, έχοντας ως είσοδο την όραση ή την ακοή και έξοδο το μυοσκελετικό σύστημα. Ως σύστημα, το REMEDES αποτελείται από έναν αριθμό ασύρματων φορητών συσκευών (Pads), οι οποίες μπορούν να τοποθετηθούν στον χώρο και να “προγραμματιστούν” ανάλογα, υλοποιώντας έτσι διάφορους τύπους ασκήσεων. Μέσω του κατάλληλου λογισμικού, για κάθε άσκηση γίνεται ανάλυση αποτελεσμάτων, ενώ παρέχονται στοιχεία επίδοσης χρήστη. Το σύστημα δίνει την δυνατότητα σύγκρισης των επιδόσεων ανάμεσα σε άλλους χρήστες ή ομάδες χρηστών. Κάθε REMEDES Pad ενεργοποιείται, παράγοντας φως συγκεκριμένου χρώματος/φωτεινότητας ή ήχο συγκεκριμένης έντασης/συχνότητας. Στη συνέχεια, ο εκάστοτε χρήστης καλείται να το “απενεργοποιήσει”, περνώντας το χέρι (ή άλλο μέλος του σώματος ανάλογα με την άσκηση) μπροστά από το εμπρόσθιο μέρος της συσκευής, οπότε και καταγράφεται με ακρίβεια ο χρόνος που πέρασε από την ενεργοποίηση έως την απενεργοποίηση του Pad. Κάθε άσκηση αποτελείται από έναν αριθμό τέτοιων ενεργοποιήσεων/απενεργοποιήσεων. Συνεπώς συνδυάζοντας διαφορετικές τοπολογίες και διαφορετικά ερεθίσματα (χρώματα, φωτεινότητες, ήχο), μπορεί να δημιουργηθεί ένα μεγάλο εύρος ασκήσεων διαφορετικής πολυπλοκότητας και δυσκολίας. Το σύστημα καταμετρά τις έγκυρες, άκυρες και εσφαλμένες απενεργοποιήσεις, όπως και όλους τους χρόνους απόκρισης, και παρουσιάζει τα αποτελέσματα σε γραφική κι επεξεργάσιμη μορφή. Ένα από τα ανταγωνιστικά πλεονεκτήματα του συστήματος REMEDES σε σχέση με άλλα, παρόμοια, συστήματα είναι ότι υποστηρίζει μέσα από τη διαδικτυακή γραφική του διεπαφή τη δημιουργία και εκτέλεση ασκήσεων τυχαίας ενεργοποίησης (όπου το σύστημα αποφασίζει ποιες συσκευές θα ενεργοποιηθούν ανάλογα με παραμέτρους εισόδου), ασκήσεις προκαθορισμένων βημάτων, όπως και ασκήσεις ελέγχου μνήμης. Στη συγκεκριμένη ομιλία θα παρουσιαστούν ο τρόπος λειτουργίας του συστήματος, οι οθόνες διεπαφής όπου εμφανίζονται τα αποτελέσματα και μία μικρή επίδειξη ενδεικτικών ασκήσεων. @conference{EmmanouilPICAD2019, | |
Konstantinos Panayiotou, Emmanouil Tsardoulias, Christopher Zolotas, Iason Paraskevopoulos, Alexandra Chatzicharistou, Alexandros Sahinis, Stathis Dimitriadis, Dimitra Ntzioni, Christopher Mpekos, Giannis Manousaridis, Aris Georgoulas and Andreas Symeonidis
"Ms Pacman and the Robotic Ghost: A Modern Cyber-Physical Remake of the Famous Pacman Game"
2019 Sixth International Conference on Internet of Things: Systems, Management and Security (IOTSMS), 2019 Oct
Robotics and Internet of Things (IoT) are two of the most blooming scientific areas during the last years. Robotics has gained a lot of attention in the last decades and includes several disciplines (mapping, localization, planning, control etc.), while IoT is a quite new and exciting area, where seamless data aggregation and resource utilization from heterogeneous physical objects (e.g. devices, sensor networks and robots) is defined via multi-layer architectures. Moreover, Cyber-Physical systems (CPS) share similar concepts and principles with the IoT, focused on interconnecting physical and computational resources via multi-layer architectures. The current paper joins the Robotics and CPS disciplines via an architecture where heterogeneous physical and computational elements exist (robots, web app, message broker etc.), so as to implement a cyber-physical port of the famous Pacman game, called RoboPacman. @conference{etsardouPacman2019, | |
Anastasios Tzitzis, Spyros Megalou, Stavroula Siachalou, Emmanouil Tsardoulias, Traianos Yioultsis and Antonis G. Dimitriou
"3D Localization of RFID Tags with a Single Antenna by a Moving Robot and ”Phase ReLock”"
2019 IEEE International Conference on RFID Technology and Applications (RFID-TA), 2019 Sep
In this paper, we propose a novel method for the three dimensional (3D) localization of RFID tags, by deploying a single RFID antenna on a robotic platform. The constructed robot is capable of performing Simultaneous Localization (of its own position) and Mapping (SLAM) of the environment and then locating the tags around its path. The proposed method exploits the unwrapped measured phase of the backscattered signal, in such manner that the localization problem can be solved rapidly by standard optimization methods. Three dimensional solution is accomplished with a single antenna on top of the robot, by forcing the robot to traverse non-straight paths (e.g. s-shaped) along the environment. It is proven theoretically and experimentally that any non-straight path reduces the locus of possible solutions to only two points along the 3D space, instead of the circle that represents the corresponding locus for typical straight robot trajectories. As a consequence, by applying our proposed method ”Phase Relock” along the known half-plane of the search-space, the unique solution is rapidly found. We experimentally compare our method against the ”holographic” method, which represents the accuracy benchmark in priorart, deploying commercial off-the-shelf (COTS) equipment. Both algorithms find the unique solution, as expected. Furthermore, ”Phase ReLock” overcomes the calculations-grid constraints of the latter. Thus, better accuracy is achieved, while, more importantly, Phase-Relock is orders of magnitude faster, allowing for the applicability of the method in real-time inventorying and localization. @conference{etsardouRfid12019, | |
Stavroula Siachalou, Spyros Megalou, Anastasios Tzitzis, Emmanouil Tsardoulias, John Sahalos, Traianos Yioultsis and Antonis G. Dimitriou
"Robotic Inventorying and Localization of RFID Tags, Exploiting Phase-Fingerprinting"
2019 IEEE International Conference on RFID Technology and Applications (RFID-TA), 2019 Sep
In this paper we investigate the performance of phase-based fingerprinting for the localization of RFID-tagged items in warehouses and large retail stores, by deploying ground and aerial RFID-equipped robots. The measured phases of the target RFID tags, collected along a given robot's trajectory, are compared to the corresponding phase-measurements of reference RFID tags; i.e. tags placed at known locations. The advantage of the method is that it doesn't need to estimate the robot's trajectory, since estimation is carried out by comparing phase measurements collected at neighboring time-intervals. This is of paramount importance for an RFID equipped drone, destined to fly indoors, since its weight should be kept as low as possible, in order to constrain its diameter correspondingly small. The phase measurements are initially unwrapped and then fingerprinting is applied. We compare the phase-fingerprinting with RSSI based fingerprinting. Phase-fingerprinting is significantly more accurate, because of the shape of the phase-function, which is typically U-shaped, with its minimum, measured at the point of the trajectory, when the robot-tag distance is minimised. Experimental accuracy of 15cm is typically achieved, depending on the density of the reference tags' grid. @conference{etsardouRfid22019, | |
Michail D. Papamichail, Themistoklis Diamantopoulos, Vasileios Matsoukas, Christos Athanasiadis and Andreas L. Symeonidis
"Towards Extracting the Role and Behavior of Contributors in Open-source Projects"
Proceedings of the 14th International Conference on Software Technologies - Volume 1: ICSOFT, pp. 536-543, SciTePress, 2019 Jul
Lately, the popular open source paradigm and the adoption of agile methodologies have changed the way soft-ware is developed. Effective collaboration within software teams has become crucial for building successful products. In this context, harnessing the data available in online code hosting facilities can help towards understanding how teams work and optimizing the development process. Although there are several approaches that mine contributions’ data, they usually view contributors as a uniform body of engineers, and focus mainlyon the aspect of productivity while neglecting the quality of the work performed. In this work, we design a methodology for identifying engineer roles in development teams and determine the behaviors that prevail for each role. Using a dataset of GitHub projects, we perform clustering against the DevOps axis, thus identifying three roles: developers that are mainly preoccupied with code commits, operations engineers that focus on task assignment and acceptance testing, and the lately popular role of DevOps engineers that are a mix of both.Our analysis further extracts behavioral patterns for each role, this way assisting team leaders in knowing their team and effectively directing responsibilities to achieve optimal workload balancing and task allocati @inproceedings{icsoft19devops, | |
Kyriakos C. Chatzidimitriou, Michail D. Papamichail, Themistoklis Diamantopoulos, Napoleon-Christos Oikonomou and Andreas L. Symeonidis
"npm Packages as Ingredients: A Recipe-based Approach - Volume 1: ICSOFT"
Proceedings of the 14th International Conference on Software Technologies, pp. 544-551, SciTePress, 2019 Jul
The sharing and growth of open source software packages in the npm JavaScript (JS) ecosystem has beenexponential, not only in numbers but also in terms of interconnectivity, to the extend that often the size of de-pendencies has become more than the size of the written code. This reuse-oriented paradigm, often attributedto the lack of a standard library in node and/or in the micropackaging culture of the ecosystem, yields interest-ing insights on the way developers build their packages. In this work we view the dependency network of thenpm ecosystem from a “culinary” perspective. We assume that dependencies are the ingredients in a recipe,which corresponds to the produced software package. We employ network analysis and information retrievaltechniques in order to capture the dependencies that tend to co-occur in the development of npm packages andidentify the communities that have been evolved as the main drivers for npm’s exponential grow. @inproceedings{icsoft19npm, | |
Maria Kotouza, Fotis Psomopoulos and Periklis A. Mitkas
New Trends in Databases and Information Systems, pp. 564-569, Springer International Publishing, Cham, 2019 Sep
Nowadays, a wide range of sciences are moving towards the Big Data era, producing large volumes of data that require processing for new knowledge extraction. Scientific workflows are often the key tools for solving problems characterized by computational complexity and data diversity, whereas cloud computing can effectively facilitate their efficient execution. In this paper, we present a generative big data analysis workflow that can provide analytics, clustering, prediction and visualization services to datasets coming from various scientific fields, by transforming input data into strings. The workflow consists of novel algorithms for data processing and relationship discovery, that are scalable and suitable for cloud infrastructures. Domain experts can interact with the workflow components, set their parameters, run personalized pipelines and have support for decision-making processes. As case studies in this paper, two datasets consisting of (i) Documents and (ii) Gene sequence data are used, showing promising results in terms of efficiency and performance. @inproceedings{Kotouza19NTDIS, | |
Ιoannis Maniadis, Konstantinos N. Vavliakis and Andreas L. Symeonidis
"Banner Personalization for e-Commerce"
AIAI 2019: Artificial Intelligence Applications and Innovations, pp. 635-646, 2019 May
@inproceedings{kvavAIAI2019, | |
Spyros Megalou, Anastasios Tzitzis, Stavroula Siachalou, Traianos Yioultsis, John Sahalos, Emmanouil Tsardoulias, Alexandros Filotheou, Andreas Symeonidis, Loukas Petrou and Antonis G. Dimitriou
"Fingerprinting Localization of RFID tags with Real-Time Performance-Assessment, using a Moving Robot"
13th European Conference of Antennas and Propagation, Krakow, Poland, 2019 Jan
@conference{Megalou2019, | |
Konstantinos Panayiotou, Emmanouil Tsardoulias, Christopher Zolotas, Iason Paraskevopoulos, Alexandra Chatzicharistou, Alexandros Sahinis, Stathis Dimitriadis, Dimitra Ntzioni, Christopher Mpekos, Giannis Manousaridis, Aris Georgoulas and Andreas L. Symeonidis
"Ms Pacman and the Robotic Ghost: A Modern Cyber-Physical Remake of the Famous Pacman Game"
2019 Sixth International Conference on Internet of Things: Systems, Management and Security (IOTSMS), pp. 147-154, 2019 Oct
@inproceedings{panayiotou2019ms, | |
Eleni Poptsi, Despoina Moraitou, Tsardoulias Emmanouil, Panayiotou Konstantinos, Symeonidis Andreas, Petrou Loukas and Magda Tsolaki
"Συστοιχία REMEDES: Ένα νέο ηλεκτρονικό εργαλείο αξιολόγησης ικανοτήτων νοητικού ελέγχου στη γήρανση"
11th Panhellenic Conference on Alzheimer's Disease & 3rd Mediterranean Conference on Neurodegenerative Diseases PICAD & MeCoND, Thessaloniki, Greece, 2019 Feb
Στις μέρες μας, υπάρχουν αρκετά νευροψυχολογικά εργαλεία που έχουν χρησιμοποιηθεί για τον διαχωρισμό των νοητικά υγιών ατόμων άνω των 65 ετών, από τα άτομα με Υποκειμενική Νοητική Δυσλειτουργία (ΥΝΔ), με Ήπια Νοητική Δυσλειτουργία (ΗΝΔ) και άνοια. Με βάση την υπάρχουσα βιβλιογραφία, οι ικανότητες νοητικού ελέγχου όπως η αναστολή και η εργαζόμενη μνήμη έχουν συσχετιστεί με νοητική έκπτωση και άνοια. Ωστόσο, οι δοκιμασίες που χρησιμοποιούνται έως σήμερα τείνουν να επηρεάζονται είτε από το εκπαιδευτικό επίπεδο του εξεταζόμενου, είτε από αντίστοιχες γλωσσικές μειονεξίες. Γι’ αυτό το λόγο τα υπάρχοντα εργαλεία φαίνεται να μην είναι ιδιαίτερα ευαίσθητα στη διαφορική διάγνωση μεταξύ των παραπάνω ομάδων. Κατά συνέπεια, η σχεδίαση κατάλληλων συστοιχιών/εργαλείων που μπορούν να εκτιμήσουν της ικανότητες νοητικού ελέγχου, χωρίς να απαιτούν γλωσσικές ικανότητες (μειώνοντας έτσι την επίδραση του εκπαιδευτικού επιπέδου των συμμετεχόντων) παραμένει ένα θέμα ιδιαίτερα επίκαιρο. Για το σκοπό αυτό δημιουργήθηκε μια συστοιχία αξιολόγησης του νοητικού ελέγχου προσαρμόζοντας το σύστημα “REMEDES1”, ένα σύστημα μέτρησης αντανακλαστικών/αντίδρασης. Η συστοιχία αυτή επικεντρώνεται σε τρεις διαφορετικές πτυχές του νοητικού ελέγχου (εργαζόμενη μνήμη, προσοχή κι εκτελεστική λειτουργία). Η πρώτη δοκιμασία εξετάζει ικανότητες εργαζόμενης μνήμης, ενώ η επόμενη εκτιμά ικανότητες εποπτικού συστήματος προσοχής. Οι τελευταία δοκιμασία διερευνά τον ανασταλτικό έλεγχο και την εναλλαγή κανόνων/έργων. Η συστοιχία δοκιμασιών REMEDES4Alzheimer θα εφαρμοστεί σε 150 συμμετέχοντες (n=150), οι οποίοι θα χωριστούν σε τέσσερις ομάδες: α) υγιείς ηλικιωμένοι, β) ηλικιωμένοι με Υποκειμενική Νοητική Διαταραχή (ΥNΔ), γ) διαγνωσθέντες με Ήπια Νοητική Διαταραχή (ΗNΔ) και δ) διαγνωσθέντες με ήπια άνοια. Στη συγκεκριμένη ομιλία θα παρουσιαστεί η φιλοσοφία και η δομή της συστοιχίας, τα πλεονεκτήματά της σε σχέση με τις υπόλοιπες συστοιχίες νοητικού ελέγχου που υπάρχουν, καθώς και τα πρώτα αποτελέσματα από το πιλοτικό στάδιο της μελέτης. @conference{PoptsiMeCoND2019, | |
Eleni Poptsi, Despoina Moraitou, Tsardoulias Emmanouil, Panayiotou Konstantinos, Symeonidis Andreas, Petrou Loukas and Magda Tsolaki
"Αξιολόγηση του νοητικού ελέγχου στη γήρανση με τη χρήση ηλεκτρονικών εργαλείων μέσω του συστήματος αντανακλαστικών/αντίδρασης REMEDES4Alzheimer"
11th Panhellenic Conference on Alzheimer's Disease & 3rd Mediterranean Conference on Neurodegenerative Diseases PICAD & MeCoND, Thessaloniki, Greece, 2019 Feb
Η συστοιχία REMEDES4Alzheimer είναι ένα νέο ηλεκτρονικό εργαλείο που στοχεύει στην αξιολόγηση ικανοτήτων νοητικού ελέγχου και απευθύνεται σε ηλικιωμένους με νοητικά ελλείμματα. Η συστοιχία αυτή αποτελεί προσαρμογή του ήδη υπάρχοντος συστήματος αντανακλαστικών/αντίδρασης REMEDES. Στόχος της παρούσας συστοιχίας είναι η διαφορική διάγνωση μεταξύ ήπιων και μείζονων νοητικών διαταραχών από το φυσιολογικό γήρας και από το φυσιολογικό γήρας με ήπια νοητικά παράπονα. Το σύστημα αποτελείται από 7 φορητές συσκευές (REMEDES pads), οι οποίες είναι προγραμματισμένες να ενεργοποιούνται, δηλαδή να παράγουν χρώμα ή/και ήχο ανάλογα με τις απαιτήσεις της εκάστοτε υποδοκιμασίας. Για τις ανάγκες της αξιολόγησης του νοητικού ελέγχου στη γήρανση έχουν προσαρτηθεί στα REMEDES pads γραφικές αναπαραστάσεις ζώων, οι οποίες συνδυάζονται με τις αντίστοιχες ηχητικές αναπαραστάσεις. Ο εξεταζόμενος καλείται να απενεργοποιήσει τα REMEDES pads, περνώντας το χέρι του πάνω από κάθε ένα, ανάλογα με τις οδηγίες της κάθε υπο-δοκιμασίας. Κατά τη διάρκεια της εκτέλεσης της συστοιχίας δοκιμασιών, οι οδηγίες που αναφέρονται στα έργα δίνονται τόσο λεκτικά όσο και μη λεκτικά (μέσω εικονικών αναπαραστάσεων-σκίτσων). Η συστοιχία περιλαμβάνει δοκιμασίες οι οποίες αξιολογούν τρεις βασικές πλευρές των ικανοτήτων νοητικού ελέγχου. Η πρώτη δοκιμασία αξιολογεί ικανότητες εργαζόμενης μνήμης και συγκεκριμένα ικανότητες αποθήκευσης, επεξεργασίας και ενημέρωσης της εργαζόμενης μνήμης. Η δεύτερη δοκιμασία αξιολογεί το εποπτικό σύστημα προσοχής και συγκεκριμένα την οπτική και ακουστική επιλεκτική προσοχή, την συντηρούμενη και διαμοιραζόμενη προσοχή. Η τρίτη και τελευταία δοκιμασία αξιολογεί εκτελεστικές ικανότητες και συγκεκριμένα τον ανασταλτικό έλεγχο, την εναλλαγή των κανόνων/έργων και τη νοητική ευελιξία. Στη συγκεκριμένη ομιλία θα παρουσιαστεί η δομή και το περιεχόμενο της κάθε δοκιμασίας, ο τρόπος βαθμολόγησης της συστοιχίας καθώς και οι δυνατότητες που δίνει το γραφικό περιβάλλον του συστήματος. @conference{PoptsiPICAD2019, | |
Christos Psarras, Themistoklis Diamantopoulos and Andreas Symeonidis
"A Mechanism for Automatically Summarizing Software Functionality from Source Code"
Proceedings of the 2019 IEEE International Conference on Software Quality, Reliability and Security (QRS), pp. 121-130, IEEE, Sofia, Bulgaria, 2019 Jul
When developers search online to find software components to reuse, they usually first need to understand the container projects/libraries, and subsequently identify the required functionality. Several approaches identify and summarize the offerings of projects from their source code, however they often require that the developer has knowledge of the underlying topic modeling techniques; they do not provide a mechanism for tuning the number of topics, and they offer no control over the top terms for each topic. In this work, we use a vectorizer to extract information from variable/method names and comments, and apply Latent Dirichlet Allocation to cluster the source code files of a project into different semantic topics.The number of topics is optimized based on their purity with respect to project packages, while topic categories are constructed to provide further intuition and Stack Exchange tags are used to express the topics in more abstract terms @inproceedings{QRS2019, | |
Stavroula Siachalou, Spyros Megalou, Anastasios Tzitzis, Emmanouil Tsardoulias, John Sahalos, Traianos Yioultsis and Antonis Dimitriou
"Robotic Inventorying and Localization of RFID Tags"
2019 IEEE International Conference on RFID Technology and Applications (RFID-TA), pp. 362-367, IEEE, 2019 Sep
In this paper we investigate the performance of phase-based fingerprinting for the localization of RFID-tagged items in warehouses and large retail stores, by deploying ground and aerial RFID-equipped robots. The measured phases of the target RFID tags, collected along a given robot’s trajectory, are compared to the corresponding phase-measurements of reference RFID tags; i.e. tags placed at known locations. The advantage of the method is that it doesn’t need to estimate the robot’s trajectory, since estimation is carried out by comparing phase measurements collected at neighboring time-intervals. This is of paramount importance for an RFID equipped drone, destined to fly indoors, since its weight should be kept as low as possible, in order to constrain its diameter correspondingly small. The phase measurements are initially unwrapped and then fingerprinting is applied. We compare the phase-fingerprinting with RSSI based fingerprinting. Phase-fingerprinting is significantly more accurate, because of the shape of the phase-function, which is typically U-shaped, with its minimum, measured at the point of the trajectory, when the robot-tag distance is minimised. Experimental accuracy of 15cm is typically achieved, depending on the density of the reference tags’ grid. @inproceedings{siachalou2019robotic, | |
Anastasios Tzitzis, Spyros Megalou, Stavroula Siachalou, Traianos Yioultsis, John Sahalos, Emmanouil Tsardoulias, Alexandros Filotheou, Andreas Symeonidis, Loukas Petrou and Antonis G. Dimitriou
"Phase ReLock - Localization of RFID Tags by a Moving Robot"
13th European Conference of Antennas and Propagation, Krakow, Poland, 2019 Jan
@conference{Tzitzis2019, | |
Anastasios Tzitzis, Spyros Megalou, Stavroula Siachalou, Emmanouil Tsardoulias, Traianos Yioultsis and Antonis Dimitriou
"3D Localization of RFID Tags with a Single Antenna by a Moving Robot and” Phase ReLock”"
2019 IEEE International Conference on RFID Technology and Applications (RFID-TA), pp. 273-278, IEEE, 2019 Sep
In this paper, we propose a novel method for the three dimensional (3D) localization of RFID tags, by deploying a single RFID antenna on a robotic platform. The constructed robot is capable of performing Simultaneous Localization (of its own position) and Mapping (SLAM) of the environment and then locating the tags around its path. The proposed method exploits the unwrapped measured phase of the backscattered signal, in such manner that the localization problem can be solved rapidly by standard optimization methods. Three dimensional solution is accomplished with a single antenna on top of the robot, by forcing the robot to traverse non-straight paths (e.g. s-shaped) along the environment. It is proven theoretically and experimentally that any non-straight path reduces the locus of possible solutions to only two points along the 3D space, instead of the circle that represents the corresponding locus for typical straight robot trajectories. As a consequence, by applying our proposed method ”Phase Relock” along the known half-plane of the search-space, the unique solution is rapidly found. We experimentally compare our method against the ”holographic” method, which represents the accuracy benchmark in priorart, deploying commercial off-the-shelf (COTS) equipment. Both algorithms find the unique solution, as expected. Furthermore, ”Phase ReLock” overcomes the calculations-grid constraints of the latter. Thus, better accuracy is achieved, while, more importantly, Phase-Relock is orders of magnitude faster, allowing for the applicability of the method in real-time inventorying and localization. @inproceedings{tzitzis20193d, | |
Konstantinos N. Vavliakis, George Katsikopoulos and Andreas L. Symeonidis
"E-commerce Personalization with Elasticsearch"
International Workshop on Web Search and Data Mining in conjunction with The 10th International Conference on Ambient Systems, Networks and Technologies (ANT 2019), Leuven, Belgium, 2019 Apr
Personalization techniques are constantly gaining traction among e-commerce retailers, since major advancements have been made at research level and the benefits are clear and pertinent. However, effectively applying personalization in real life is a challenging task, since the proper mixture of technology, data and content is complex and differs between organizations. In fact, personalization applications such as personalized search remain largely unfulfilled, especially by small and medium sized retailers, due to time and space limitations. In this paper we propose a novel approach for near real-time personalized e-commerce search that provides improved personalized results within the limited accepted time frames required for online browsing. We propose combining features such as product popularity, user interests, and query-product relevance with collaborative filtering, and implement our solution in Elasticsearch in order to achieve acceptable execution timings. We evaluate our approach against a publicly available dataset, as well as a running e-commerce store. @inproceedings{VavliakisWSDM2018, |
2018
Conference Papers
Eleni Nisioti, Kyriakos C. Chatzidimitriou and Andreas L. Symeonidis
"ICML 2018 AutoML WorkshopPredicting hyperparameters from meta-features in binary classification problems"
AutoML, http://assets.ctfassets.net/c5lel8y1n83c/5uAPDjSvcseoko2cCcQcEi/8bd1d8e3630e246946feac86271fe03b/PPC17-automl2018.pdf, Stockholm, Sweden, 2018 Jul
The presence of computationally demanding problems and the current inability to auto-matically transfer experience from the application of past experiments to new ones delaysthe evolution of knowledge itself. In this paper we present the Automated Data Scientist1,a system that employs meta-learning for hyperparameter selection and builds a rich ensem-ble of models through forward model selection in order to automate binary classificationtasks. Preliminary evaluation shows that the system is capable of coping with classificationproblems of medium complexity. @conference{2018Nisioti, | |
Sotirios-Filippos Tsarouchis, Maria Th. Kotouza, Fotis E. Psomopoulos and Pericles A. Mitkas
"A Multi-metric Algorithm for Hierarchical Clustering of Same-Length Protein Sequences"
IFIP International Conference on Artificial Intelligence Applications and Innovations, pp. 189-199, Springer, Cham, 2018 May
The identification of meaningful groups of proteins has always been a major area of interest for structural and functional genomics. Successful protein clustering can lead to significant insight, assisting in both tracing the evolutionary history of the respective molecules as well as in identifying potential functions and interactions of novel sequences. Here we propose a clustering algorithm for same-length sequences, which allows the construction of subset hierarchy and facilitates the identification of the underlying patterns for any given subset. The proposed method utilizes the metrics of sequence identity and amino-acid similarity simultaneously as direct measures. The algorithm was applied on a real-world dataset consisting of clonotypic immunoglobulin (IG) sequences from Chronic lymphocytic leukemia (CLL) patients, showing promising results. @inproceedings{2018Tsarouchis, | |
Kyriakos C. Chatzidimitriou, Michail Papamichail, Themistoklis Diamantopoulos, Michail Tsapanos and Andreas L. Symeonidis
"npm-miner: An Infrastructure for Measuring the Quality of the npm Registry"
MSR ’18: 15th International Conference on Mining Software Repositories, pp. 4, ACM, Gothenburg, Sweden, 2018 May
As the popularity of the JavaScript language is constantly increasing, one of the most important challenges today is to assess the quality of JavaScript packages. Developers often employ tools for code linting and for the extraction of static analysis metrics in order to assess and/or improve their code. In this context, we have developed npn-miner, a platform that crawls the npm registry and analyzes the packages using static analysis tools in order to extract detailed quality metrics as well as high-level quality attributes, such as maintainability and security. Our infrastructure includes an index that is accessible through a web interface, while we have also constructed a dataset with the results of a detailed analysis for 2000 popular npm packages. @inproceedings{Chatzidimitriou2018MSR, | |
Themistoklis Diamantopoulos, Georgios Karagiannopoulos and Andreas Symeonidis
"CodeCatch: Extracting Source Code Snippets from Online Sources"
IEEE/ACM 6th International Workshop on Realizing Artificial Intelligence Synergies in Software Engineering (RAISE), pp. 21-27, https://dl.acm.org/ft_gateway.cfm?id=3194107&ftid=1982571&dwn=1&CFID=87644405&CFTOKEN=833260e7cb501a7d-48967D35-AFC5-4678-82812B13D64D3DD3, 2018 May
https://dl.acm.org/ft_gateway.cfm?id=3194107&ftid=1982571&dwn=1&CFID=87644405&CFTOKEN=833260e7cb501a7d-48967D35-AFC5-4678-82812B13D64D3DD3 @inproceedings{Diamantopoulos2018, | |
Anastasios Dimanidis, Kyriakos C. Chatzidimitriou and Andreas L. Symeonidis
"A Natural Language Driven Approach for Automated Web API Development: Gherkin2OAS"
WWW ’18 Companion: The 2018 Web Conference Companion, pp. 6, Lyon, France, 2018 Apr
Speeding up the development process of Web Services, while adhering to high quality software standards is a typical requirement in the software industry. This is why industry specialists usually suggest \\"driven by\\" development approaches to tackle this problem. In this paper, we propose such a methodology that employs Specification Driven Development and Behavior Driven Development in order to facilitate the phases of Web Service requirements elicitation and specification. Furthermore, we introduce gherkin2OAS, a software tool that aspires to bridge the aforementioned development approaches. Through the suggested methodology and tool, one may design and build RESTful services fast, while ensuring proper functionality. @inproceedings{Dimanidis2018, | |
Maria Th. Kotouza, Konstantinos N. Vavliakis, Fotis E. Psomopoulos and Pericles A. Mitkas
"A Hierarchical Multi-Metric Framework for Item Clustering"
5th International Conference on Big Data Computing Applications and Technologies, pp. 191-197, IEEE/ACM, Zurich, Switzerland, 2018 Dec
Item clustering is commonly used for dimensionality reduction, uncovering item similarities and connections, gaining insights of the market structure and recommendations. Hierarchical clustering methods produce a hierarchy structure along with the clusters that can be useful for managing item categories and sub-categories, dealing with indirect competition and new item categorization as well. Nevertheless, baseline hierarchical clustering algorithms have high computational cost and memory usage. In this paper we propose an innovative scalable hierarchical clustering framework, which overcomes these limitations. Our work consists of a binary tree construction algorithm that creates a hierarchy of the items using three metrics, a) Identity, b) Similarity and c) Entropy, as well as a branch breaking algorithm which composes the final clusters by applying thresholds to each branch of the tree. ?he proposed framework is evaluated on the popular MovieLens 20M dataset achieving significant reduction in both memory consumption and computational time over a baseline hierarchical clustering algorithm. @inproceedings{KotouzaVPM18, | |
Panagiotis G. Mousouliotis, Konstantinos L. Panayiotou, Emmanouil G. Tsardoulias, Loukas P. Petrou and Andreas L. Symeonidis
"Expanding a robots life: Low power object recognition via FPGA-based DCNN deployment"
MOCAST, https://arxiv.org/abs/1804.00512, 2018 Mar
FPGAs are commonly used to accelerate domain-specific algorithmic implementations, as they can achieve impressive performance boosts, are reprogrammable and exhibit minimal power consumption. In this work, the SqueezeNet DCNN is accelerated using an SoC FPGA in order for the offered object recognition resource to be employed in a robotic application. Experiments are conducted to investigate the performance and power consumption of the implementation in comparison to deployment on other widely-used computational systems. thanks you! @conference{Mousouliotis2018, | |
Michail Papamichail, Themistoklis Diamantopoulos, Ilias Chrysovergis, Philippos Samlidis and Andreas Symeonidis
Proceedings of the 2018 Workshop on Machine Learning Techniques for Software Quality Evaluation (MaLTeSQuE), https://www.researchgate.net/publication/324106989_User-Perceived_Reusability_Estimation_based_on_Analysis_of_Software_Repositories, 2018 Mar
The popularity of open-source software repositories has led to a new reuse paradigm, where online resources can be thoroughly analyzed to identify reusable software components. Obviously, assessing the quality and specifically the reusability potential of source code residing in open software repositories poses a major challenge for the research community. Although several systems have been designed towards this direction, most of them do not focus on reusability. In this paper, we define and formulate a reusability score by employing information from GitHub stars and forks, which indicate the extent to which software components are adopted/accepted by developers. Our methodology involves applying and assessing different state-of-the-practice machine learning algorithms, in order to construct models for reusability estimation at both class and package levels. Preliminary evaluation of our methodology indicates that our approach can successfully assess reusability, as perceived by developers. @inproceedings{Papamichail2018MaLTeSQuE, | |
Emmanouil G. Tsardoulias, Konstantinos L. Panayiotou, Christoforos Zolotas, Alexandros Philotheou, Anreas L. Symeonidis and Loukas Petrou
"From classical to cloud robotics: Challenges and potential"
3rd International Workshop on Microsystems, Sindos Campus, ATEI Thessaloniki, Greece, 2018 Dec
Nowadays, a rapid transition from the classical robotic systems to more modern concepts like Cloud or IoT robotics is being experienced. The current paper briefly overviews the benefits robots can have, as parts of the increasingly interconnected world. @conference{TsardouliasMicrosystems2018, | |
Konstantinos N. Vavliakis, Maria Th. Kotouza, Andreas L. Symeonidis and Pericles A. Mitkas
"Recommendation Systems in a Conversational Web"
Proceedings of the 14th International Conference on Web Information Systems and Technologies - Volume 1: WEBIST,, pp. 68-77, SciTePress, 2018 Jan
In this paper we redefine the concept of Conversation Web in the context of hyper-personalization. We argue that hyper-personalization in the WWW is only possible within a conversational web where websites and users continuously “discuss” (interact in any way). We present a modular system architecture for the conversational WWW, given that adapting to various user profiles and multivariate websites in terms of size and user traffic is necessary, especially in e-commerce. Obviously there cannot be a unique fit-to-all algorithm, but numerous complementary personalization algorithms and techniques are needed. In this context, we propose PRCW, a novel hybrid approach combining offline and online recommendations using RFMG, an extension of RFM modeling. We evaluate our approach against the results of a deep neural network in two datasets coming from different online retailers. Our evaluation indicates that a) the proposed approach outperforms current state-of-art methods in small-medium datasets and can improve performance in large datasets when combined with other methods, b) results can greatly vary in different datasets, depending on size and characteristics, thus locating the proper method for each dataset can be a rather complex task, and c) offline algorithms should be combined with online methods in order to get optimal results since offline algorithms tend to offer better performance but online algorithms are necessary for exploiting new users and trends that turn up. @conference{webist18, |
2017
Conference Papers
Maria Th. Kotouza, Antonios C. Chrysopoulos and Pericles A. Mitkas
"Segmentation of Low Voltage Consumers for Designing Individualized Pricing Policies"
European Energy Market (EEM), 2017 14th International Conference, pp. 1-6, IEEE, Dresden, Germany, 2017 Jun
In recent years, the Smart Grid paradigm has opened a vast set of opportunities for all participating parties in the Energy Markets (i.e. producers, Distribution and Transmission System Operators, retailers, consumers), providing two-way data communication, increased security and grid stability. Furthermore, the liberation of distribution and energy services has led towards competitive Energy Market environments [4]. In order to maintain their existing customers\' satisfaction level high, as well as reaching out to new ones, suppliers must provide better and more reliable energy services, that are specifically tailored to each customer or to a group of customers with similar needs. Thus, it is necessary to identify segments of customers that have common energy characteristics via a process called Consumer Load Profiling (CLP) [16]. @inproceedings{2017Kotouza, | |
Panagiotis Doxopoulos, Konstantinos Panayiotou, Emmanouil Tsardoulias and Andreas L. Symeonidis
"Creating an extrovert robotic assistant via IoT networking devices"
International Conference on Cloud and Robotics, Saint Quentin, France, 2017 Nov
The communication and collaboration of Cyber-Physical Systems, including machines and robots, among themselves and with humans, is expected to attract researchers\\' interest for the years to come. A key element of the new revolution is the Internet of Things (IoT). IoT infrastructures enable communication between different connected devices using internet protocols. The integration of robots in an IoT platform can improve robot capabilities by providing access to other devices and resources. In this paper we present an IoT-enabled application including a NAO robot which can communicate through an IoT platform with a reflex measurement system and a hardware node that provides robotics-oriented services in the form of RESTful web services. An activity reminder application is also included, illustrating the extension capabilities of the system. @inproceedings{Doxopoulos2017, | |
Valasia Dimaridou, Alexandros-Charalampos Kyprianidis, Michail Papamichail, Themistoklis Diamantopoulos and Andreas Symeonidis
"Towards Modeling the User-perceived Quality of Source Code using Static Analysis Metrics"
Proceedings of the 12th International Conference on Software Technologies - Volume 1: ICSOFT, pp. 73-84, SciTePress, 2017 Jul
Nowadays, software has to be designed and developed as fast as possible, while maintaining quality standards. In this context, developers tend to adopt a component-based software engineering approach, reusing own implementations and/or resorting to third-party source code. This practice is in principle cost-effective, however it may lead to low quality software products. Thus, measuring the quality of software components is of vital importance. Several approaches that use code metrics rely on the aid of experts for defining target quality scores and deriving metric thresholds, leading to results that are highly context-dependent and subjective. In this work, we build a mechanism that employs static analysis metrics extracted from GitHub projects and defines a target quality score based on repositories’ stars and forks, which indicate their adoption/acceptance by the developers’ community. Upon removing outliers with a one-class classifier, we employ Principal Feature Analysis and exam ine the semantics among metrics to provide an analysis on five axes for a source code component: complexity, coupling, size, degree of inheritance, and quality of documentation. Neural networks are used to estimate the final quality score given metrics from all of these axes. Preliminary evaluation indicates that our approach can effectively estimate software quality. @inproceedings{icsoft17, | |
Emmanouil Krasanakis, Eleftherios Spyromitros-Xioufis, Symeon Papadopoulos and Yiannis Kompatsiaris
"Tunable Plug-In Rules with Reduced Posterior Certainty Loss in Imbalanced Datasets"
Proceedings of the First International Workshop on Learning with Imbalanced Domains: Theory and Applications, pp. 116-128, PMLR, ECML-PKDD, Skopje, Macedonia, 2017 Sep
Classifiers have difficulty recognizing under-represented minorities in imbalanced datasets, due to their focus on minimizing the overall misclassification error. This introduces predictive biases against minority classes. Post-processing plug-in rules are popular for tackling class imbalance, but they often affect the certainty of base classifier posteriors, when the latter already perform correct classification. This shortcoming makes them ill-suited to scoring tasks, where informative posterior scores are required for human interpretation. To this end, we propose the ILoss metric to measure the impact of imbalance-aware classifiers on the certainty of posterior distributions. We then generalize post-processing plug-in rules in an easily tunable framework and theoretically show that this framework tends to improve performance balance. Finally, we experimentally assert that appropriate usage of our framework can reduce ILoss while yielding similar performance, with respect to common imbalance-aware measures, to existing plug-in rules for binary problems. @inproceedings{Krasanakis2017, | |
Konstantinos Panayiotou, Sofia E. Reppou, George Karagiannis, Emmanouil Tsardoulias, Aristeidis G. Thallas and Andreas L. Symeonidis
"Robotic applications towards an interactive alerting system for medical purposes"
30th IEEE International Symposium on Computer-Based Medical Systems (IEEE CBMS), Thessaloniki, 2017 Jan
Social consumer robots are slowly but strongly invading our everyday lives as their prices are becoming lower and lower, constituting them affordable for a wide range of civilians. There has been a lot of research concerning the potential applications of social robots, some of which may implement companionship or proxying technology-related tasks and assisting in everyday household endeavors, among others. In the current work, the RAPP framework is being used towards easily creating robotic applications suitable for utilization as a socially interactive alerting system with the employment of the NAO robot. The developed application stores events in an on-line calendar, directly via the robot or indirectly via a web environment, and asynchronously informs an end-user of imminent events @inproceedings{Panayiotou2017, | |
Vasilis N. Remmas, Konstantinos Panayiotou, Emmanouil Tsardoulias and Andreas L. Symeonidis
"SRCA - The Scalable Robotic Cloud Agents Architecture"
International Conference on Cloud and Robotics, Saint Quentin, France, 2017 Nov
In an effort to penetrate the market at an affordable cost, consumer robots tend to provide limited processing capabilities, just enough to serve the purpose they have been designed for. However, a robot, in principle, should be able to interact and process the constantly increasing information streams generated from sensors or other devices. This would require the implementation of algorithms and mathematical models for the accurate processing of data volumes and significant computational resources. It is clear that as the data deluge continues to grow exponentially, deploying such algorithms on consumer robots will not be easy. Current work presents a cloud-based architecture that aims to offload computational resources from robots to a remote infrastructure, by utilizing and implementing cloud technologies. This way robots are allowed to enjoy functionality offered by complex algorithms that are executed on the cloud. The proposed system architecture allows developers and engineers not specialised in robotic implementation environments to utilize generic robotic algorithms and services off-the-shelf. @inproceedings{Remmas2017, |
2016
Conference Papers
Kyriakos Chatzidimitriou, Konstantinos Vavliakis, Andreas L. Symeonidis and Pericles A. Mitkas
"Towards defining the structural properties of efficient consumer social networks on the electricity grid"
AI4SG SETN Workshop on AI for the Smart Grid, 2016 May
Energy markets have undergone important changes at the conceptual level over the last years. Decentralized supply, small-scale pro- duction, smart grid optimization and control are the new building blocks. These changes o er substantial opportunities for all energy market stake- holders, some of which however, remain largely unexploited. Small-scale consumers as a whole, account for signi cant amount of energy in current markets (up to 40%). As individuals though, their consumption is triv- ial, and their market power practically non-existent. Thus, it is necessary to assist small-scale energy market stakeholders, combine their market power. Within the context of this work, we propose Consumer Social Networks (CSNs) as a means to achieve the objective. We model con- sumers and present a simulation environment for the creation of CSNs and provide a proof of concept on how CSNs can be formulated based on various criteria. We also provide an indication on how demand response programs designed based on targeted incentives may lead to energy peak reductions. @conference{2016ChatzidimitriouSETN, | |
Themistoklis Diamantopoulos, Klearchos Thomopoulos and Andreas L. Symeonidis
"QualBoa: Reusability-aware Recommendations of Source Code Components"
IEEE/ACM 13th Working Conference on Mining Software Repositories, 2016 May
Contemporary software development processes involve finding reusable software components from online repositories and integrating them to the source code, both to reduce development time and to ensure that the final software project is of high quality. Although several systems have been designed to automate this procedure by recommending components that cover the desired functionality, the reusability of these components is usually not assessed by these systems. In this work, we present QualBoa, a recommendation system for source code components that covers both the functional and the quality aspects of software component reuse. Upon retrieving components, QualBoa provides a ranking that involves not only functional matching to the query, but also a reusability score based on configurable thresholds of source code metrics. The evaluation of QualBoa indicates that it can be effective for recommending reusable source code. @conference{2016DiamantopoulosIEEE/ACM, | |
Themistoklis Diamantopoulos, Antonis Noutsos and Andreas L. Symeonidis
"DP-CORE: A Design Pattern Detection Tool for Code Reuse"
6th International Symposium on Business Modeling and Software Design (BMSD), -, Rhodes, Greece, 2016 Dec
In order to maintain, extend or reuse software projects one has to primarily understand what a system does and how well it does it. And, while in some cases information on system functionality exists, information covering the non-functional aspects is usually unavailable. Thus, one has to infer such knowledge by extracting design patterns directly from the source code. Several tools have been developed to identify design patterns, however most of them are limited to compilable and in most cases executable code, they rely on complex representations, and do not offer the developer any control over the detected patterns. In this paper we present DP-CORE, a design pattern detection tool that defines a highly descriptive representation to detect known and define custom patterns. DP-CORE is flexible, identifying exact and approximate pattern versions even in non-compilable code. Our analysis indicates that DP-CORE provides an efficient alternative to existing design pattern detection tools. @conference{2016DiamantopoulosSBMSD, | |
Michail Papamichail, Themistoklis Diamantopoulos and Andreas L. Symeonidis
"User-Perceived Source Code Quality Estimation based on Static Analysis Metrics"
2016 IEEE International Conference on Software Quality, Reliability and Security (QRS), Vienna, Austria, 2016 Aug
The popularity of open source software repositories and the highly adopted paradigm of software reuse have led to the development of several tools that aspire to assess the quality of source code. However, most software quality estimation tools, even the ones using adaptable models, depend on fixed metric thresholds for defining the ground truth. In this work we argue that the popularity of software components, as perceived by developers, can be considered as an indicator of software quality. We present a generic methodology that relates quality with source code metrics and estimates the quality of software components residing in popular GitHub repositories. Our methodology employs two models: a one-class classifier, used to rule out low quality code, and a neural network, that computes a quality score for each software component. Preliminary evaluation indicates that our approach can be effective for identifying high quality software components in the context of reuse. @inproceedings{2016PapamichailIEEE, | |
Fotis Psomopoulos, Athanassios Kintsakis and Pericles Mitkas
"A pan-genome approach and application to species with photosynthetic capabilities"
15th European Conference on Computational Biology, The Hague, Netherlands, 2016 Sep
The abundance of genome data being produced by the new sequencing techniques is providing the opportunity to investigate gene diversity at a new level. A pan-genome analysis can provide the framework for estimating the genomic diversity of the data set at hand and give insights towards the understanding of its observed characteristics. Currently, there exist several tools for pan-genome studies, mostly focused on prokaryote genomes and their respective attributes. Here we provide a systematic approach for constructing the groups inherently associated with a pan-genome analysis, using the complete proteome data of photosynthetic genomes as the driving case study. As opposed to similar studies, the presented method requires a complete information system (i.e. complete genomes) in order to produce meaningful results. The method was applied to 95 genomes with photosynthetic capabilities, including cyanobacteria and green plants, as retrieved from UniProt and Plaza. Due to the significant computational requirements of the analysis, we utilized the Federated Cloud computing resources provided by the EGI infrastructure. The analysis ultimately produced 37,680 protein families, with a core genome comprising of 102 families. An investigation of the families’ distribution revealed two underlying but expected subsets, roughly corresponding to bacteria and eukaryotes. Finally, an automated functional annotation of the produced clusters, through assignment of PFAM domains to the participating protein sequences, allowed the identification of the key characteristics present in the core genome, as well as of selected multi-member families. @inproceedings{2016PsomopoulosECCB, | |
Emmanouil Stergiadis, Athanassios Kintsakis, Fotis Psomopoulos and Pericles A. Mitkas
"A scalable Grid Computing framework for extensible phylogenetic profile construction"
12th International Conference on Artificial Intelligence Applications and Innovations, pp. 455-462, 12th International Conference on Artificial Intelligence Applications and Innovations, Thessaloniki, Greece, September, 2016 Sep
Current research in Life Sciences without doubt has been established as a Big Data discipline. Beyond the expected domain-specific requirements, this perspective has put scalability as one of the most crucial aspects of any state-of-the-art bioinformatics framework. Sequence alignment and construction of phylogenetic profiles are common tasks evident in a wide range of life science analyses as, given an arbitrary big volume of genomes, they can provide useful insights on the functionality and relationships of the involved entities. This process is often a computational bottleneck in existing solutions, due to its inherent complexity. Our proposed distributed framework manages to perform both tasks with significant speed-up by employing Grid Computing resources provided by EGI in an efficient and optimal manner. The overall workflow is both fully automated, thus making it user friendly, and fully detached from the end-users terminal, since all computations take place on Grid worker nodes. @inproceedings{2016Stergiadis, | |
Aristeidis Thallas, Emmanouil Tsardoulias and Loukas Petrou
"Particle Filter - Scan Matching SLAM Recovery Under Kinematic Model Failures"
2016 24th Mediterranean Conference on Control and Automation (MED), 2016 Jun
Two of the most predominant approaches regarding the SLAM problem are the Rao-Blackwellized particle filters and the Scan Matching algorithms, each approach presenting its own deficiencies. In particular, particle filters suffer from potential particle impoverishment, whereas lack of environmental features can cause scan matching methods to collapse. In the current paper a multi-threaded combination of Rao-Blackwellized particle filters with a scan matching algorithm (CRSM SLAM) aiming to overcome those defects, whilst exploiting each method's advantages is presented. CRSM is employed in feature-rich environments while concurrently reducing the particle filter dispersion, whilst the particle filter allows the maintenance of the correct hypothesis in environments with scarcity of information. Finally, a method to reduce the number of particle filter resamplings, employing topological information is proposed. @conference{etsardouMed12016, | |
Aristeidis Thallas, Emmanouil Tsardoulias and Loukas Petrou
"Particle Filter - Scan Matching Hybrid SLAM Employing Topological Information"
2016 24th Mediterranean Conference on Control and Automation (MED), 2016 Jun
Two of the most predominant approaches regarding the SLAM problem are the Rao-Blackwellized particle filters and the Scan Matching algorithms, each approach presenting its own deficiencies. In particular, particle filters suffer from potential particle impoverishment, whereas lack of environmental features can cause scan matching methods to collapse. In the current paper a multi-threaded combination of Rao-Blackwellized particle filters with a scan matching algorithm (CRSM SLAM) aiming to overcome those defects, whilst exploiting each method's advantages is presented. CRSM is employed in feature-rich environments while concurrently reducing the particle filter dispersion, whilst the particle filter allows the maintenance of the correct hypothesis in environments with scarcity of information. Finally, a method to reduce the number of particle filter resamplings, employing topological information is proposed. @conference{etsardouMed22016, | |
Aristeidis G. Thallas, Konstantinos Panayiotou, Emmanouil Tsardoulias, Andreas L. Symeonidis, Pericles A. Mitkas and George G. Karagiannis
"Relieving robots from their burdens: The Cloud Agent concept"
2016 5th IEEE International Conference on Cloud Networking (Cloudnet), 2016 Oct
The consumer robotics concept has already invaded our everyday lives, however two major drawbacks have become apparent both for the roboticists and the consumers. The first is that these robots are pre-programmed to perform specific tasks and usually their software is proprietary, thus not open to "interventions". The second is that even if their software is open source, low-cost robots usually lack sufficient resources such as CPU power or memory capabilities, thus forbidding advanced algorithms to be executed in-robot. Within the context of RAPP (Robotic Applications for Delivering Smart User Empowering Applications) we treat robots as platforms, where applications can be downloaded and automatically deployed. Furthermore, we propose and implement a novel multi-agent architecture, empowering robots to offload computations in entities denoted as Cloud Agents. This paper discusses the respective architecture in detail. @conference{etsardouRobotBurden2016, |
2015
Conference Papers
Themistoklis Diamantopoulos and Andreas Symeonidis
"Employing Source Code Information to Improve Question-Answering in Stack Overflow"
The 12th Working Conference on Mining Software Repositories (MSR 2015), pp. 454-457, Florence, Italy, 2015 May
Nowadays, software development has been greatlyinfluenced by question-answering communities, such as Stack Overflow. A new problem-solving paradigm has emerged, as developers post problems they encounter that are then answered by the community. In this paper, we propose a methodology that allows searching for solutions in Stack Overflow, using the main elements of a question post, including not only its title, tags, and body, but also its source code snippets. We describe a similarity scheme for these elements and demonstrate how structural information can be extracted from source code snippets and compared to further improve the retrieval of questions. The results of our evaluation indicate that our methodology is effective on recommending similar question posts allowing community members to search without fully forming a question @conference{2015DiamantopoulosMSR, | |
Themistoklis Diamantopoulos and Andreas Symeonidis
"Towards Interpretable Defect-Prone Component Analysis using Genetic Fuzzy Systems"
IEEE/ACM 4th International Workshop on Realizing Artificial Intelligence Synergies in Software Engineering (RAISE, pp. 32-38, Florence, Italy, 2015 May
The problem of Software Reliability Prediction is attracting the attention of several researchers during the last few years. Various classification techniques are proposed in current literature which involve the use of metrics drawn from version control systems in order to classify software components as defect-prone or defect-free. In this paper, we create a novel genetic fuzzy rule-based system to efficiently model the defect-proneness of each component. The system uses a Mamdani-Assilian inference engine and models the problem as a one-class classification task. System rules are constructed using a genetic algorithm, where each chromosome represents a rule base (Pittsburgh approach). The parameters of our fuzzy system and the operators of the genetic algorithm are designed with regard to producing interpretable output. Thus, the output offers not only effective classification, but also a comprehensive set of rules that can be easily visualized to extract useful conclusions about the metrics of the software. @inproceedings{2015DiamantopoulosRAISE, | |
Athanassios M. Kintsakis, Antonios Chysopoulos and Pericles Mitkas
"Agent-based short-term load and price forecasting using a parallel implementation of an adaptive PSO-trained local linear wavelet neural network"
European Energy Market (EEM), pp. 1 - 5, 2015 May
Short-Term Load and Price forecasting are crucial to the stability of electricity markets and to the profitability of the involved parties. The work presented here makes use of a Local Linear Wavelet Neural Network (LLWNN) trained by a special adaptive version of the Particle Swarm Optimization algorithm and implemented as parallel process in CUDA. Experiments for short term load and price forecasting, up to 24 hours ahead, were conducted for energy market datasets from Greece and the USA. In addition, the fast response time of the system enabled its encapsulation in a PowerTAC agent, competing in a real time environment. The system displayed robust all-around performance in a plethora of real and simulated energy markets, each characterized by unique patterns and deviations. The low forecasting error, real time performance and the significant increase in the profitability of an energy market agent show that our approach is a powerful prediction tool, with multiple expansion possibilities. @conference{2015KintsakisEEM, | |
Pericles A. Mitkas
"Assistive Robots as Future Caregivers: The RAPP Approach"
Automation Conference, 2015 Mar
As our societies are affected by a dramatic demographic change, the percentage of elderly and people requiring support in their daily life is expected to increase in the near future and caregivers will not be enough to assist and support them. Socially interactive robots can help confront this situation not on- ly by physically assisting people but also by functioning as a companion. The rising sales figures of robots point towards a trend break concerning robotics. To lower the cost for developers and to increase their interest in developing ro- botic applications, the RAPP approach introduces the idea of robots as plat- forms. RAPP (A Software Platform for Delivering Smart User Empowering Robotic Applications) aims to provide a software platform in order to support the creation and delivery of robotic applications (RApps) targeting people at risk of exclusion, especially older people. The open-source software platform will provide an API with the required functionality for the implementation of RApps. It will also provide access to the robots’ sensors and actuators employ- ing higher level commands, by adding a middleware stack with functionalities suitable for different kinds of robots. RAPP will expand the robots’ computa- tional and storage capabilities and enable machine learning operations, distri- buted data collection and processing. Through a special repository for RApps, the platform will support knowledge sharing among robots in order to provide personalized applications based on adaptation to individuals. The use of a common API will facilitate the development of improved applications deploya- ble for a variety of robots. These applications target people with different needs, capabilities and expectations, while at the same time respect their privacy and autonomy. The RAPP approach can lower the cost of robotic applications de- velopment and it is expected to have a profound effect in the robotics market @conference{2015MitkasACRAPP, | |
Fotis Psomopoulos, Olga Vrousgou and Pericles A. Mitkas
"Large-scale modular comparative genomics: the Grid approach"
23rd Annual International Conference on Intelligent Systems for Molecular Biology (ISMB) / 14th European Conference on Computational Biology (ECCB), 2015 Jul
@conference{2015PsomopoulosAICISMB, | |
Alexandros Gkiokas, Emmanouil G. Tsardoulias and Pericles A. Mitkas
"Hive Collective Intelligence for Cloud Robotics A Hybrid Distributed Robotic Controller Design for Learning and Adaptation"
Automation Conference, 2015 Mar
The recent advent of Cloud Computing, inevitably gave rise to Cloud Robotics. Whilst the field is arguably still in its infancy, great promise is shown regarding the problem of limited computational power in Robotics. This is the most evident advantage of Cloud Robotics, but, other much more significant yet subtle advantages can now be identified. Moving away from traditional Robotics, and approaching Cloud Robotics through the prism of distributed systems or Swarm Intelligence offers quite an interesting composure; physical robots deployed across different areas, may delegate tasks to higher intelligence agents residing in the cloud. This design has certain distinct attributes, similar with the organisation of a Hive or bee colony. Such a parallelism is crucial for the foundations set hereinafter, as they express through the hive design, a new scheme of distributed robotic architectures. Delegation of agent intelligence, from the physical robot swarms to the cloud controllers, creates a unique type of Hive Intelligence, where the controllers residing in the cloud, may act as the brain of a ubiquitous group of robots, whilst the robots themselves act as proxies for the Hive Intelligence. The sensors of the hive system providing the input and output are the robots, yet the information processing may take place collectively, individually or on a central hub, thus offering the advantages of a hybrid swarm and cloud controller. The realisation that radical robotic architectures can be created and implemented with current Artificial Intelligence models, raises interesting questions, such as if robots belonging to a hive, can perform tasks and procedures better or faster, and if can they learn through their interactions, and hence become more adaptive and intelligent. @conference{2015TsardouliasHCIAC, | |
Emmanouil G. Tsardoulias, C Zielinski, Wlodzimierz Kasprzak, Sofia Reppou, Andreas L. Symeonidis, Pericles A. Mitkas and George Karagiannis
"Merging Robotics and AAL ontologies: The RAPP methodology"
Automation Conference, 2015 Mar
Cloud robotics is becoming a trend in the modern robotics field, as it became evident that true artificial intelligence can be achieved only by sharing collective knowledge. In the ICT area, the most common way to formulate knowledge is via the ontology form, where different meanings connect semantically. Additionally, there is a considerable effort to merge robotics with assisted living concepts, as the modern societies suffer from lack of caregivers for the persons in need. In the current work, an attempt is performed to merge a robotic and an AAL ontology, as well as utilize it in the RAPP Project (EU-FP7). @conference{2015TsardouliasMRALL, | |
Tsardoulias, E. G., Andreas Symeonidis and and P. A. Mitkas.
"An automatic speech detection architecture for social robot oral interaction"
In Proceedings of the Audio Mostly 2015 on Interaction With Sound, p. 33. ACM, Island of Rhodes, 2015 Oct
Social robotics have become a trend in contemporary robotics research, since they can be successfully used in a wide range of applications. One of the most fundamental communication skills a robot must have is the oral interaction with a human, in order to provide feedback or accept commands. And, although text-to-speech is an almost solved problem, this isn @conference{2015TsardouliasPAMIWS, | |
Konstantinos Vavliakis, Anthony Chrysopoulos, Kyriakos C. Chatzidimitriou, Andreas L. Symeonidis and Pericles A. Mitkas
"CASSANDRA: a simulation-based, decision-support tool for energy market stakeholders"
SimuTools, 2015 Dec
Energy gives personal comfort to people, and is essential for the generation of commercial and societal wealth. Nevertheless, energy production and consumption place considerable pressures on the environment, such as the emission of greenhouse gases and air pollutants. They contribute to climate change, damage natural ecosystems and the man-made environment, and cause adverse e ects to human health. Lately, novel market schemes emerge, such as the formation and operation of customer coalitions aiming to improve their market power through the pursuit of common bene ts.In this paper we present CASSANDRA, an open source1,expandable software platform for modelling the demand side of power systems, focusing on small scale consumers. The structural elements of the platform are a) the electrical installations (i.e. households, commercial stores, small industries etc.), b) the respective appliances installed, and c) the electrical consumption-related activities of the people residing in the installations.CASSANDRA serves as a tool for simulation of real demandside environments providing decision support for energy market stakeholders. The ultimate goal of the CASSANDRA simulation functionality is the identi cation of good practices that lead to energy eciency, clustering electric energy consumers according to their consumption patterns, and the studying consumer change behaviour when presented with various demand response programs. @conference{2015VavliakisSimuTools, | |
Olga Vrousgou, Fotis Psomopoulos and Pericles Mitkas
"A grid-enabled modular framework for efficient sequence analysis workflows"
16th International Conference on Engineering Applications of Neural Network, Island of Rhodes, 2015 Oct
In the era of Big Data in Life Sciences, efficient processing and analysis of vast amounts of sequence data is becoming an ever daunting challenge. Among such analyses, sequence alignment is one of the most commonly used procedures, as it provides useful insights on the functionality and relationship of the involved entities. Sequence alignment is one of the most common computational bottlenecks in several bioinformatics workflows. We have designed and implemented a time-efficient distributed modular application for sequence alignment, phylogenetic profiling and clustering of protein sequences, by utilizing the European Grid Infrastructure. The optimal utilization of the Grid with regards to the respective modules, allowed us to achieve significant speedups to the order of 1400%. @conference{2015VrousgouICEANN, | |
Christoforos Zolotas and Andreas Symeonidis
"Towards an MDA Mechanism for RESTful Services Development"
The 18th International Conference on Model Driven Engineering Languages and Systems, Ottawa, Canada, 2015 Oct
—Automated software engineering research aspiresto lead to more consistent software, faster delivery and lowerproduction costs. Meanwhile, RESTful design is rapidly gainingmomentum towards becoming the primal software engineeringparadigm for the web, due to its simplicity and reusability. Thispaper attempts to couple the two perspectives and take the firststep towards applying the MDE paradigm to RESTful servicedevelopment at the PIM zone. A UML profile is introduced,which performs PIM meta-modeling of RESTful web servicesabiding by the third level of Richardson’s maturity model. Theprofile embeds a slight variation of the MVC design pattern tocapture the core REST qualities of a resource. The proposedprofile is followed by an indicative example that demonstrateshow to apply the concepts presented, in order to automate PIMproduction of a system according to MOF stack. Next stepsinclude the introduction of the corresponding CIM, PSM andcode production.Index Terms—Model Driven Engineering; RESTful services;UML Profiles; Meta-modeling; Automated Software Engineering @conference{2015ZolotasICMDELS, |
2014
Conference Papers
Christos Dimou, Fani Tzima, Andreas L. Symeonidis and and Pericles A. Mitkas
"Performance Evaluation of Agents and Multi-agent Systems using Formal Specifications in Z Notation"
Lecture Notes on Agents and Data Mining Interaction, pp. 50-54, Springer, Baltimore, Maryland, USA, 2014 May
Software requirements are commonlywritten in natural language, making themprone to ambiguity, incompleteness and inconsistency. By converting requirements to formal emantic representations, emerging problems can be detected at an early stage of the development process, thus reducing the number of ensuing errors and the development costs. In this paper, we treat the mapping from requirements to formal representations as a semantic parsing task. We describe a novel data set for this task that involves two contributions: first, we establish an ontology for formally representing requirements; and second, we introduce an iterative annotation scheme, in which formal representations are derived through step-wise refinements. @inproceedings{2014Dimou, | |
Rafaila Grigoriou and Andreas L. Symeonidis
"Towards the Design of User Friendly Search Engines for Software Projects"
Lecture Notes on Natural Language Processing and Information Systems, pp. 164-167, Springer International Publishing, Chicago, Illinois, 2014 Jun
Robots are fast becoming a part of everyday life. This rise can be evidenced both through the public news and announcements, as well as in recent literature in the robotics scientific communities. This expanding development requires new paradigms in producing the necessary software to allow for the users @inproceedings{2014GrigoriouTDUFSESP, | |
Fotis Psomopoulos, Emmanouil Tsardoulias, Alexandros Giokas, Cezary Zielinski, Vincent Prunet, Ilias Trochidis, David Daney, Manuel Serrano, Ludovic Courtes, Stratos Arampatzis and Pericles A. Mitkas
"RAPP System Architecture, Assistance and Service Robotics in a Human Environment"
International Conference on Intelligent Robots and Systems (IEEE/RSJ), Chicago, Illinois, 2014 Sep
Robots are fast becoming a part of everyday life. This rise can be evidenced both through the public news and announcements, as well as in recent literature in the robotics scientific communities. This expanding development requires new paradigms in producing the necessary software to allow for the users @conference{2014PsomopoulosIEEE/RSJ, | |
Michael Roth, Themistoklis Diamantopoulos, Ewan Klein and Andreas L. Symeonidis
"Software Requirements: A new Domain for Semantic Parsers"
Proceedings of the ACL 2014 Workshop on Semantic Parsing (SP14), pp. 50-54, Baltimore, Maryland, USA, 2014 Jun
Software requirements are commonlywritten in natural language, making themprone to ambiguity, incompleteness and inconsistency. By converting requirements to formal emantic representations, emerging problems can be detected at an early stage of the development process, thus reducing the number of ensuing errors and the development costs. In this paper, we treat the mapping from requirements to formal representations as a semantic parsing task. We describe a novel data set for this task that involves two contributions: first, we establish an ontology for formally representing requirements; and second, we introduce an iterative annotation scheme, in which formal representations are derived through step-wise refinements. @inproceedings{roth2014software, |
2013
Conference Papers
Kyriakos C. Chatzidimitriou, Konstantinos N. Vavliakis, Andreas Symeonidis and Pericles Mitkas
"Redefining the market power of small-scale electricity consumers through consumer social networks"
10th IEEE International Conference on e-Business Engineering (ICEBE 2013), pp. 30-44, Springer Berlin Heidelberg, 2013 Jan
136 @inproceedings{2013ChatzidimitriouICEBE, | |
Antonios Chrysopoulos, Christos Diou, Andreas L. Symeonidis and Pericles Mitkas
"Agent-based small-scale energy consumer models for energy portfolio management"
Proceedings of the 2013 IEEE/WIC/ACM International Conference on Intelligent Agent Technology (IAT 2013), pp. 45-50, Atlanta, GA, USA, 2013 Jan
Locating software bugs is a difficult task, especially if they do not lead to crashes. Current research on automating non-crashing bug detection dictates collecting function call traces and representing them as graphs, and reducing the graphs before applying a subgraph mining algorithm. A ranking of potentially buggy functions is derived using frequency statistics for each node (function) in the correct and incorrect set of traces. Although most existing techniques are effective, they do not achieve scalability. To address this issue, this paper suggests reducing the graph dataset in order to isolate the graphs that are significant in localizing bugs. To this end, we propose the use of tree edit distance algorithms to identify the traces that are closer to each other, while belonging to different sets. The scalability of two proposed algorithms, an exact and a faster approximate one, is evaluated using a dataset derived from a real-world application. Finally, although the main scope of this work lies in scalability, the results indicate that there is no compromise in effectiveness. @inproceedings{2013ChrysopoulosIAT, | |
Themistoklis Diamantopoulos and Andreas L. Symeonidis
"Towards Scalable Bug Localization using the Edit Distance of Call Traces"
The Eighth International Conference on Software Engineering Advances (ICSEA 2013), pp. 45-50, Venice, Italy, 2013 Oct
Locating software bugs is a difficult task, especially if they do not lead to crashes. Current research on automating non-crashing bug detection dictates collecting function call traces and representing them as graphs, and reducing the graphs before applying a subgraph mining algorithm. A ranking of potentially buggy functions is derived using frequency statistics for each node (function) in the correct and incorrect set of traces. Although most existing techniques are effective, they do not achieve scalability. To address this issue, this paper suggests reducing the graph dataset in order to isolate the graphs that are significant in localizing bugs. To this end, we propose the use of tree edit distance algorithms to identify the traces that are closer to each other, while belonging to different sets. The scalability of two proposed algorithms, an exact and a faster approximate one, is evaluated using a dataset derived from a real-world application. Finally, although the main scope of this work lies in scalability, the results indicate that there is no compromise in effectiveness. @inproceedings{2013DiamantopoulosICSEA, | |
Tsardoulias, E. G., A. Iliakopoulou, A. Kargakos, and L. Petrou
"On Global Path Planning for Occupancy Grid Maps"
22nd International Workshop on Robotics in Alpe-Adria-Danube Region, 2013 Sep
This paper considers the problem of robot path planning in indoors environments. Several approaches to tackle this problem have been proposed, which employ structures such as graphs or trees to direct robot’s movement throughout space. The current document constitutes a survey of eight well-known path planning methods, aiming at comparing and evaluating their performances in various environments of different characteristics. @conference{etsardouRaad2013, |
2012
Conference Papers
Georgios T. Andreou, Andreas L. Symeonidis, Christos Diou, Pericles A. Mitkas and Dimitrios P. Labridis
"A framework for the implementation of large scale Demand Response"
Smart Grid Technology, Economics and Policies (SG-TEP), 2012 International Conference on, Nuremberg, Germany, 2012 Jan
Agent autonomy is strongly related to learning and adaptation. Machine learning models generated, either by off-line or on-line adaptation, through the use of historical data or current environmental signals, provide agents with the necessary decision-making and generalization capabilities in competitive, dynamic, partially observable and stochastic environments. In this work, we discuss learning and adaptation in the context of the TAC SCM game. We apply a variety of machine learning and computational intelligence methods for generating the most efficient sales component of the agent, dealing with customer orders and production throughput. Along with utility maximization and bid acceptance probability estimation methods, we evaluate regression trees, particle swarm optimization, heuristic control and policy search via adaptive function approximation in order to build an efficient, near-real time, bidding mechanism. Results indicate that a suitable reinforcement learning setup coupled with the power of adaptive function approximation techniques adjusted to the problem at hand, is a good candidate for enabling high performance strategies. @inproceedings{2012andreouSGTEP2012, | |
Kyriakos C. Chatzidimitriou, Konstantinos Vavliakis, Andreas L. Symeonidis and Pericles A. Mitkas
"Policy Search through Adaptive Function Approximation for Bidding in TAC SCM"
Joint Workshop on Trading Agents Design and Analysis and Agent Mediated Electronic Commerce, 2012 May
Agent autonomy is strongly related to learning and adaptation. Machine learning models generated, either by off-line or on-line adaptation, through the use of historical data or current environmental signals, provide agents with the necessary decision-making and generalization capabilities in competitive, dynamic, partially observable and stochastic environments. In this work, we discuss learning and adaptation in the context of the TAC SCM game. We apply a variety of machine learning and computational intelligence methods for generating the most efficient sales component of the agent, dealing with customer orders and production throughput. Along with utility maximization and bid acceptance probability estimation methods, we evaluate regression trees, particle swarm optimization, heuristic control and policy search via adaptive function approximation in order to build an efficient, near-real time, bidding mechanism. Results indicate that a suitable reinforcement learning setup coupled with the power of adaptive function approximation techniques adjusted to the problem at hand, is a good candidate for enabling high performance strategies. @inproceedings{2012ChatzidimitriouAMEC, | |
Themistoklis Mavridis and Andreas L. Symeonidis
"Identifying webpage Semantics for Search Engine Optimization"
Paper presented at the 8th International Conference on Web Information Systems and Technologies (WEBIST), pp. 18-21, Porto, Portugal, 2012 Jun
The added-value of search engines is, apparently, undoubted. Their rapid evolution over the last decade has transformed them into the most important source of information and knowledge. From the end user side, search engine success implies correct results in fast and accurate manner, while also ranking of search results on a given query has to be directly correlated to the user anticipated response. From the content providers side (i.e. websites), better ranking in a search engine result set implies numerous advantages like visibility, visitability, and profit. This is the main reason for the flourishing of Search Engine Optimization (SEO) techniques, which aim towards restructuring or enriching website content, so that optimal ranking of websites in relation to search engine results is feasible. SEO techniques are becoming more and more sophisticated. Given that internet marketing is extensively applied, prior quality factors prove insufficient, by themselves, to boost ranking and the improvement of the quality of website content is also introduced. Current paper discusses such a SEO mechanism. Having identified that semantic analysis has not been widely applied in the field of SEO, a semantic approach is adopted, which employs Latent Dirichlet Allocation techniques coupled with Gibbs Sampling in order to analyze the results of search engines based on given keywords. Within the context of the paper, the developed SEO mechanism LDArank is presented, which evaluates query results through state-of-the-art SEO metrics, analyzes results content and extracts new, optimized content. @inproceedings{2012MavridisWEBIST, | |
Athanasios Papadopoulos, Konstantinos Toumpas, Antonios Chrysopoulos and Pericles A. Mitkas
"Exploring Optimization Strategies in Board Game Abalone for Alpha-Beta Seach"
IEEE Conference on Computational Intelligent and Games (CIG), pp. 63-70, Granada, Spain, 2012 Sep
This paper discusses the design and implementation of a highly efficient MiniMax algorithm for the game Abalone.For perfect information games with relatively low branching factor for their decision tree (such as Chess, Checkers etc.) anda highly accurate evaluation function, Alpha-Beta search proved to be far more efficient than Monte Carlo Tree Search. In recentyears many new techniques have been developed to improve the efficiency of the Alpha-Beta tree, applied to a variety of scientific fields. This paper explores several techniques for increasing the efficiency of Alpha-Beta Search on the board game of Abalone while introducing some new innovative techniques that proved to be very effective. The main idea behind them is the incorporation of probabilistic features to the otherwise deterministic Alpha-Beta search. @inproceedings{2012PapadopoulosCIG, | |
Andreas Symeonidis, Panagiotis Toulis and Pericles A. Mitkas
"Supporting Agent-Oriented Software Engineering for Data Mining Enhanced Agent Development"
Agents and Data Mining Interaction workshop (ADMI 2012), at the 2012 Conference on Autonimous Agents and Multiagent Systems (AAMAS), Valencia, Spain, 2012 Jun
The emergence of Multi-Agent systems as a software paradigm that most suitably fits all types of problems and architectures is already experiencing significant revisions. A more consistent approach on agent programming, and the adoption of Software Engineering standards has indicated the pros and cons of Agent Technology and has limited the scope of the, once considered, programming ‘panacea’. Nowadays, the most active area of agent development is by far that of intelligent agent systems, where learning, adaptation, and knowledge extraction are at the core of the related research effort. Discussing knowledge extraction, data mining, once infamous for its application on bank processing and intelligence agencies, has become an unmatched enabling technology for intelligent systems. Naturally enough, a fruitful synergy of the aforementioned technologies has already been proposed that would combine the benefits of both worlds and would offer computer scientists with new tools in their effort to build more sophisticated software systems. Current work discusses Agent Academy, an agent toolkit that supports: a) rapid agent application development and, b) dynamic incorporation of knowledge extracted by the use of data mining techniques into agent behaviors in an as much untroubled manner as possible. @inproceedings{2012SymeonidisADMI, | |
Konstantinos N. Vavliakis, Georgios T. Karagiannis and Periklis A. Mitkas
"Semantic Web in Cultural Heritage After 2020"
What will the Semantic Web look like 10 Years From Now? Workshop held in conjunction with the 11th International Semantic Web Conference 2012 (ISWC 2012), Boston, USA, 2012 Nov
In this paper we present the current status of semantic data management in the cultural heritage field and we focus on the challenges imposed by the multidimensionality of the information in this domain. We identify current shortcomings, thus needs, that should be addressed in the coming years to enable the integration and exploitation of the rich information deriving from the multidisciplinary analysis of cultural heritage objects, monuments and sites. Our goal is to disseminate the needsof the cultural heritage community and drive Semantic web research towards these directions. @inproceedings{2012VavliakisISWC, | |
Konstantinos N. Vavliakis, Fani A. Tzima and Pericles A. Mitkas
"Event Detection via LDA for the MediaEval2012 SED Task"
Working Notes Proceedings of the MediaEval 2012, Santa Corce in Fossabanda, Pisa, Italy, 2012 Oct
In this paper we present our methodology for the Social Event Detection Task of the MediaEval 2012 BenchmarkingInitiative. We adopt topic discovery using Latent Dirichlet Allocation (LDA), city classification using TF-IDF analysis, and other statistical and natural language processing methods. After describing the approach we employed, we present the corresponding results, and discuss the problems we faced, as well as the conclusions we drew. @inproceedings{2012VavliakisLDA, | |
Dimitrios M. Vitsios, Fotis E. Psomopoulos, Pericles A. Mitkas and Chistos A. Ouzounis
"Mutli-gemone Core Pathway Identification Through Gene Clustering"
1st Workshop on Algorithms for Data and Text Mining in Bionformatics (WADTMB 2012) in conjunction with the 8th AIAI, Halkidiki, Greece, 2012 Sep
In the wake of gene-oriented data analysis in large-scale bioinformatics studies, focus in research is currently shifting towards the analysis of the functional association of genes, namely the metabolic pathways in which genes participate. The goal of this paper is to attempt to identify the core genes in a specific pathway, based on a user-defined selection of genomes. To this end, a novel methodology has been developed that uses data from the KEGG database, and through the application of the MCL clustering algorithm, identifies clusters that correspond to different “layers” of genes, either on a phylogenetic or a functional level. The algorithm’s complexity, evaluated experimentally, is presented and the results on a characteristic case study are discussed. @inproceedings{2012VitsiosWADTMB, |
2011
Conference Papers
Zinovia Alepidou, Konstantinos N. Vavliakis and Pericles A. Mitkas
"A Semantic Tag Recommendation Framework for Collaborative Tagging Systems"
Proceedings of the Third IEEE International Conference on Social Computing, pp. 633-636, Cambridge, MA, USA, 2011 Oct
In this work we focus on folksonomies. Our goal is to develop techniques that coordinate information processing, by taking advantage of user preferences, in order to automatically produce semantic tag recommendations. To this end, we propose a generalized tag recommendation framework that conveys the semantics of resources according to different user pro?les. We present the integration of various models that take into account content, historic values, user preferences and tagging behavior to produce accurate personalized tag recommendations. Based on this information we build several Bayesian models, we evaluate their performance, and we dis-cuss differences in accuracy with respect to semantic matching criteria, and other approaches. @inproceedings{2011AlepidouSocialCom, | |
Kyriakos C. Chatzidimitriou, Ioannis Partalas, Pericles A. Mitkas and Ioannis Vlahavas
"Transferring Evolved Reservoir Features in Reinforcement Learning Tasks"
European Workshop on Reinforcement Learning, pp. 213-224, Springer Berlin Heidelberg, Athens, Greece, 2011 Sep
The major goal of transfer learning is to transfer knowledge acquired on a source task in order to facilitate learning on another, different, but usually related, target task. In this paper, we are using neuroevolution to evolve echo state networks on the source task and transfer the best performing reservoirs to be used as initial population on the target task. The idea is that any non-linear, temporal features, represented by the neurons of the reservoir and evolved on the source task, along with reservoir properties, will be a good starting point for a stochastic search on the target task. In a step towards full autonomy and by taking advantage of the random and fully connected nature of echo state networks, we examine a transfer method that renders any inter-task mappings of states and actions unnecessary. We tested our approach and that of inter-task mappings in two RL testbeds: the mountain car and the server job scheduling domains. Under various setups the results we obtained in both cases are promising. @inproceedings{2011Chatzidimitriou, | |
Andreas L. Symeonidis, Vasileios P. Gountis and Georgios T. Andreou
"A Software Agent Framework for exploiting Demand-side Consumer Social Networks in Power Systems"
Paper presented at the IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology, pp. 30--33, Lyon, France, 2011 Aug
This work introduces Energy City, a multi-agent framework designed and developed in order to simulate the power system and explore the potential of Consumer Social Networks (CSNs) as a means to promote demand-side response and raise social awareness towards energy consumption. The power system with all its involved actors (Consumers, Producers, Electricity Suppliers, Transmission and Distribution Operators) and their requirements are modeled. The semantic infrastructure for the formation and analysis of electricity CSNs is discussed, and the basic consumer attributes and CSN functionality are identified. Authors argue that the formation of such CSNs is expected to increase the electricity consumer market power by enabling them to act in a collective way. @inproceedings{2011SymeonidisICWEBIIAT, | |
Michael Tsapanos, Kiriakos C. Chatzidimitriou and Pericles A. Mitkas
"Combining Zeroth-Level Classifier System and Eligibility Traces for Real Time Strategy Games"
IEEE/WIC/ACM International Conference on Web Intelligent and Intelligent Agent Technology (WI-IAT'11), pp. 244-247, Lyons, France, 2011 Aug
This work introduces Energy City, a multi-agent framework designed and developed in order to simulate the power system and explore the potential of Consumer Social Networks (CSNs) as a means to promote demand-side response and raise social awareness towards energy consumption. The power system with all its involved actors (Consumers, Producers, Electricity Suppliers, Transmission and Distribution Operators) and their requirements are modeled. The semantic infrastructure for the formation and analysis of electricity CSNs is discussed, and the basic consumer attributes and CSN functionality are identified. Authors argue that the formation of such CSNs is expected to increase the electricity consumer market power by enabling them to act in a collective way. @inproceedings{2011TsapanosIEEE, | |
Michalis Tsapanos, Kyriakos C. Chatzidimitriou and Pericles A. Mitkas
"A Zeroth-Level Classifier System for Real Time Strategy Games"
Web Intelligence and Intelligent Agent Technology (WI-IAT), 2011 IEEE/WIC/ACM International Conference, pp. 244-247, Springer Berlin Heidelberg, Lyons, France, 2011 Aug
Real Time Strategy games (RTS) provide an interesting test bed for agents that use Reinforcement Learning (RL) algorithms. From an agent @conference{2011TsapanosWI-IAT, | |
Iraklis Tsekourakis and Andreas L. Symeonidis
"Dealing with Trust and Reputation in unreliable Multi-agent Trading Environments"
Paper presented at the 2011 Workshop on Trading Agent Design and Analysis (IJCAI 2011), pp. 21-28, Barcelona, Spain, 2011 Aug
In shared competitive environments, where information comes from various sources, agents may interact with each other in a competitive manner in order to achieve their individual goals. Numerous research efforts exist, attempting to define protocols, rules and interfaces for agents to abide by and ensure trustworthy exchange of information. Auction environments and e-commerce platforms are such paradigms, where trust and reputation are vital factors determining agent strategy. And though the process is always secured with a number of safeguards, there is always the issue of unreliability. In this context, the Agent Reputation and Trust (ART) testbed has provided researchers with the ability to test different trust and reputation strategies, in various types of trust/reputation environments. Current work attempts to identify the most viable trust and reputation models stated in the literature, while it further elaborates on the issue by proposing a robust trust and reputation mechanism. This mechanism is incorporated in our agent, HerculAgent, and tested in a variety of environments against the top performing agents of the ART competition. The paper provides a thorough analysis of ART, presents HerculAgent s architecture and dis-cuss its performance. @inproceedings{2011TsekourakisIJCAI, | |
Kyriakos C. Chatzidimitriou, Antonios C. Chrysopoulos, Andreas L. Symeonidis and Pericles A. Mitkas
"Enhancing Agent Intelligence through Evolving Reservoir Networks for Prediction in Power Stock Markets"
Agent and Data Mining Interaction 2011 Workshop held in conjuction with the conference on Autonomous Agents and Multi-Agent Systems (AAMAS) 2011, pp. 228-247, 2011 Apr
In recent years, Time Series Prediction and clustering have been employed in hyperactive and evolving environments -where temporal data play an important role- as a result of the need for reliable methods to estimate and predict the pattern or behavior of events and systems. Power Stock Markets are such highly dynamic and competitive auction environments, additionally perplexed by constrained power laws in the various stages, from production to transmission and consumption. As with all real-time auctioning environments, the limited time available for decision making provides an ideal testbed for autonomous agents to develop bidding strategies that exploit time series prediction. Within the context of this paper, we present Cassandra, a dynamic platform that fosters the development of Data-Mining enhanced Multi-agent systems. Special attention was given on the efficiency and reusability of Cassandra, which provides Plug-n-Play capabilities, so that users may adapt their solution to the problem at hand. Cassandra’s functionality is demonstrated through a pilot case, where autonomously adaptive Recurrent Neural Networks in the form of Echo State Networks are encapsulated into Cassandra agents, in order to generate power load and settlement price prediction models in typical Day-ahead Power Markets. The system has been tested in a real-world scenario, that of the Greek Energy Stock Market. @inproceedings{2012ChatzidimitriouAAMAS, | |
Kyriakos C. Chatzidimitriou, Lampros C. Stavrogiannis, Andreas Symeonidis and Pericles A. Mitkas
"An Adaptive Proportional Value-per-Click Agent for Bidding in Ad Auctions"
Trading Agent Design and Analysis (TADA) 2011 Workshop held in conjuction with the International Joint Conference on Artificial Intelligence (IJCAI) 2011, pp. 21-28, Barcelona, Spain, 2011 Jul
Sponsored search auctions constitutes the most important source of revenue for search engine companies, offering new opportunities for advertisers. The Trading Agent Competition (TAC) Ad Auctions tournament is one of the first attempts to study the competition among advertisers for their placement in sponsored positions along with organic search engine results. In this paper, we describe agent Mertacor, a simulation-based game theoretic agent coupled with on-line learning techniques to optimize its behavior that successfully competed in the 2010 tournament. In addition, we evaluate different facets of our agent to draw conclusions about certain aspects of its strategy. @inproceedings{Chatzidimitriou2011, | |
Dimitrios Vitsios, Fotis E. Psomopoulos, Pericles A. Mitkas and Christos A. Ouzounis
"Detecting Species Evolution Through Metabolic Pathways"
6th Conference of the Hellenic Society for computational Biology & Bioinformatics (HSCBB11), pp. 16, Patra, Greece, 2011 Oct
The emergence and evolution of metabolic pathways represented a crucial step in molecular and cellular evolution. Withthe current advances in genomics and proteomics, it has become imperative to explore the impact of gene evolution as reflected in the metabolic signature of each genome (Zhang et al. (2006)). To this end a methodology is presented, which applies a clustering algorithm to genes from different species participating in the same pathway. @inproceedings{PsomopoulosHSCBB11, | |
Konstantinos N. Vavliakis, Konstantina Gemenetzi and Pericles A. Mitkas
"A correlation analysis of web social media"
Proceedings of the International Conference on Web Intelligence, Mining and Semantics, pp. 54:1--54:5, ACM, Songdal, Norway, 2011 Jan
In this paper we analyze and compare three popular content creation and sharing websites, namely Panoramio, YouTube and Epinions. This analysis aims in advancing our understanding of Web Social Media and their impact, and may be useful in creating feedback mechanisms for increasing user participation and sharing. For each of the three websites, we select ?ve fundamental factors appearing in all content centered Web Social Media and we use regression analysis to calculate their correlation. We present findings of statistically important correlations among these key factors and we rank the discovered correlations according to the degree of their in?uence. Furthermore, we perform analysis of variance in distinct subgroups of the collected data and we discuss differences found in the characteristics of these subgroups and how these differences may affect correlation results. Although we acknowledge that correlation does not imply causality, the discovered correlations may be a ?rst step towards discovering causality laws behind content contribution, commenting and the formulation of friendship relations. These causality laws are useful for boosting the user participation in social media @inproceedings{Vavliakis:2011:CAW:1988688.1988752, |
2010
Conference Papers
Kyriakos C. Chatzidimitriou and Pericles A. Mitkas
"A NEAT Way for Evolving Echo State Networks"
European Conference on Artificial Intelligence, pp. 909-914, IOS Press, Alexandroupoli, Greece, 2010 Aug
The Reinforcement Learning (RL) paradigm is an appropriateformulation for agent, goal-directed, sequential decisionmaking. In order though for RL methods to perform well in difficult,complex, real-world tasks, the choice and the architecture ofan appropriate function approximator is of crucial importance. Thiswork presents a method of automatically discovering such functionapproximators, based on a synergy of ideas and techniques that areproven to be working on their own. Using Echo State Networks(ESNs) as our function approximators of choice, we try to adaptthem, by combining evolution and learning, for developing the appropriatead-hoc architectures to solve the problem at hand. Thechoice of ESNs was made for their ability to handle both non-linearand non-Markovian tasks, while also being capable of learning online,through simple gradient descent temporal difference learning.For creating networks that enable efficient learning, a neuroevolutionprocedure was applied. Appropriate topologies and weights wereacquired by applying the NeuroEvolution of Augmented Topologies(NEAT) method as a meta-search algorithm and by adaptingideas like historical markings, complexification and speciation, to thespecifics of ESNs. Our methodology is tested on both supervised andreinforcement learning testbeds with promising results. @inproceedings{2010ChatzidimitriouECAI, | |
Kyriakos C. Chatzidimitriou, Fotis E. Psomopoulos and Pericles A. Mitkas
"Grid-enabled parameter initialization for high performance machine learning tasks"
5th EGEE User Forum, pp. 113-114, 2010 Apr
In this work we use the NeuroEvolution of augmented Topologies (NEAT) methodology, for optimising Echo State Networks (ESNs), in order to achieve high performance in machine learning tasks. The large parameter space of NEAT, the many variations of ESNs and the stochastic nature of enolutionary computation, requiring manyevaluations for staatistically valid conclusions, promotes the Grid as a a viable solution for robustly evaluationg the alternatives and deriving significant conclusions. @inproceedings{2010ChatzidimitriouEGEEForum, | |
Nausheen S. Khuram, Andreas L. Symeonidis and Awais Majeed
"Wage – A Web Service- and Agent-based Generic Auctioning Environment"
Paper presented at the 2010 IADIS International Conference on Intelligent Systems and Agents, Freiburg, Germany, 2010 Jul
@inproceedings{2010KhuramISA, | |
Pericles A. Mitkas
"From Theory and the Research Lav to an Innocative Product for the Greek and the International Market: Agent MerTACor"
1st Private Equity Forum, Transforming the Crisis to Opportunities for Greece, Athens, Greece, 2010 Oct
During the last decade, there has been intense research and development in creating methodologies and tools able to map Relational Databases with the Resource Description Framework. Although some systems have gained wider acceptance in the Semantic Web community, they either require users to learn a declarative language for encoding mappings, or have limited expressivity. Thereupon we present RDOTE, a framework for easily transporting data residing in Relational Databases into the Semantic Web. RDOTE is available under GNU/GPL license and provides friendly graphical interfaces, as well as enough expressivity for creating custom RDF dumps. @inproceedings{2010MitkasTCOG10, | |
Fotis E. Psomopoulos and Pericles A. Mitkas
"Multi Level Clustering of Phylogenetic Profiles"
BioInformatics and BioEngineering (BIBE), 2010 IEEE International Conference, pp. 308-309, Freiburg, Germany, 2010 May
The prediction of gene function from genome sequences is one of the main issues in Bioinformatics. Most computational approaches are based on the similarity between sequences to infer gene function. However, the availability of several fully sequenced genomes has enabled alternative approaches, such as phylogenetic profiles. Phylogenetic profiles are vectors which indicate the presence or absence of a gene in other genomes. The main concept of phylogenetic profiles is that proteins participating in a common structural complex or metabolic pathway are likely to evolve in a correlated fashion. In this paper, a multi level clustering algorithm of phylogenetic profiles is presented, which aims to detect inter- and intra-genome gene clusters. @conference{2010PsomopoulosBIBE, | |
Fotis E. Psomopoulos, Pericles A. Mitkas and Christos A. Ouzounis
"Clustering of discrete and fuzzy phylogenetic profiles"
5th Conference of the Hellenic Society For Computational Biology and Bioinformatics - HSCBB, pp. 58, Alexandroupoli, Greece, 2010 Oct
Phylogenetic profiles have long been a focus of interest in computational genomics. Encoding the subset of organisms that contain a homolog of a gene or protein, phylogenetic profiles are originally defined as binary vectors of n entries, where n corresponds to the number of target genomes. It is widely accepted that similar profiles especially those not connected by sequence similarity correspond to a correlated pattern of functional linkage. To this end, our study presents two methods of phylogenetic profile data analysis, aiming at detecting genes with peculiar, unique characteristics. Genes with similar phylogenetic profiles are likely to have similar structure or function, such as participating to a common structural complex or to a common pathway. Our two methods aim at detecting those outlier profiles of “interesting†genes, or groups of genes, with different characteristics from their parent genome. @inproceedings{2010PsomopoulosHSCBB, | |
Andreas L. Symeonidis and Pericles A. Mitkas
"Monitoring Agent Communication in Soft Real-Time Environments"
Paper presented at the IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology, pp. 265--268, Los Alamitos, CA, USA, 2010 Jan
Real-time systems can be defined as systems operating under specific timing constraints, either hard or soft ones. In principle, agent systems are considered inappropriate for such kinds of systems, due to the asynchronous nature of their communication protocols, which directly influences their temporal behavior. Nevertheless, multi-agent systems could be successfully employed for solving problems where failure to meet a deadline does not have serious consequences, given the existence of a fail-safe system mechanism. Current work focuses on the analysis of multi-agent systems behavior under such soft real-time constraints. To this end, ERMIS has been developed: an integrated framework that provides the agent developer with the ability to benchmark his/her own architecture and identify its limitations and its optimal timing behavior, under specific hardware/software constraints. A variety of MAS configurations have been tested and indicative results are discussed within the context of this paper. @inproceedings{2010SymeonidisWIIAT, | |
Fani A. Tzima, Fotis E. Psomopoulos and Pericles A. Mitkas
"An investigation of the effect of clustering-based initialization on Learning Classifiers Systems"
5th EGEE User Forum, pp. 111-112, 2010 Apr
Strength-based Learning Classifier Systems (LCS) are machine learning systems designed to tackle both sequential and single-step decision tasks by coupling a gradually evolving population of rules with a reinforcement component. ZCS-DM, a Zeroth-level Classifier System for Data Mining, is a novel algorithm in this field, recently shown to be very effective in several benchmark classification problems. In this paper, we evaluate the effect of clustering-based initialization on the algorithm’s performance, utilizing the EGEE infrastructure as a robust framework for an efficient parameter sweep. @inproceedings{2010TzimaEGEEForum, | |
Konstantinos N. Vavliakis, Theofanis K Grollios and Pericles A. Mitkas
"RDOTE - Transforming Relational Databases into Semantic Web Data"
9th International Semantic Web Conference (ISWC2010), 2010 Nov
During the last decade, there has been intense research and development in creating methodologies and tools able to map Relational Databases with the Resource Description Framework. Although some systems have gained wider acceptance in the Semantic Web community, they either require users to learn a declarative language for encoding mappings, or have limited expressivity. Thereupon we present RDOTE, a framework for easily transporting data residing in Relational Databases into the Semantic Web. RDOTE is available under GNU/GPL license and provides friendly graphical interfaces, as well as enough expressivity for creating custom RDF dumps. @inproceedings{2010Vavliakis-ISWC, | |
Konstantinos N. Vavliakis, Theofanis K. Grollios and Pericles A. Mitkas
"R. - Transforming Relational Databases into Semantic Web Data"
International Semantic Web Conference, 2010 Jan
@inproceedings{2010VavliakisISWC, | |
Kyriakos C. Chatzidimitriou, Konstantinos N. Vavliakis, Andreas L. Symeonidis and Pericles A. Mitkas
"Towards Understanding How Personality, Motivation, and Events Trigger Web User Activity"
Web Intelligence and Intelligent Agent Technology, IEEE/WIC/ACM International Conference, pp. 615-618, IEEE Computer Society, Los Alamitos, CA, USA, 2010 Jan
Web 2.0 provided internet users with a dynamic medium, where information is updated continuously and anyone can participate. Though preliminary anal-ysis exists, there is still little understanding on what exactly stimulates users to actively participate, create and share content in online communities. In this paper we present a methodology that aspires to identify and analyze those events that trigger web user activity, content creation and sharing in Web 2.0. Our approach is based on user personality and motiva-tion, and on the occurrence of events with a personal or global impact. The proposed methodology was ap-plied on data collected from Flickr and analysis was performed through the use of statistics and data mining techniques. @inproceedings{2010VavliakisWI, |
2009
Conference Papers
Antonios C. Chrysopoulos, Andreas L. Symeonidis and Pericles A. Mitkas
"Improving agent bidding in Power Stock Markets through a data mining enhanced agent platform"
Agents and Data Mining Interaction workshop AAMAS 2009, pp. 111-125, Springer-Verlag, Budapest, Hungary, 2009 May
Like in any other auctioning environment, entities participating in Power Stock Markets have to compete against other in order to maximize own revenue. Towards the satisfaction of their goal, these entities (agents - human or software ones) may adopt different types of strategies - from naive to extremely complex ones - in order to identify the most profitable goods compilation, the appropriate price to buy or sell etc, always under time pressure and auction environment constraints. Decisions become even more difficult to make in case one takes the vast volumes of historical data available into account: goods\\\\92 prices, market fluctuations, bidding habits and buying opportunities. Within the context of this paper we present Cassandra, a multi-agent platform that exploits data mining, in order to extract efficient models for predicting Power Settlement prices and Power Load values in typical Day-ahead Power markets. The functionality of Cassandra is discussed, while focus is given on the bidding mechanism of Cassandra\\\\92s agents, and the way data mining analysis is performed in order to generate the optimal forecasting models. Cassandra has been tested in a real-world scenario, with data derived from the Greek Energy Stock market. @inproceedings{2009ChrysopoulosADMI, | |
Anthonios C. Chrysopoulos, Andreas L. Symeonidis and Pericles A. Mitkas
"Creating and Reusing Metric Graphs for Evaluating Agent Performance in the Supply Chain Management Domain"
Third Electrical and Computer Engineering Department Student Conference, pp. 245-267, IGI Global, Thessaloniki, Greece, 2009 Apr
The scope of this chapter is the presentation of Data Mining techniques for knowledge extraction in proteomics, taking into account both the particular features of most proteomics issues (such as data retrieval and system complexity), and the opportunities and constraints found in a Grid environment. The chapter discusses the way new and potentially useful knowledge can be extracted from proteomics data, utilizing Grid resources in a transparent way. Protein classification is introduced as a current research issue in proteomics, which also demonstrates most of the domain – specific traits. An overview of common and custom-made Data Mining algorithms is provided, with emphasis on the specific needs of protein classification problems. A unified methodology is presented for complex Data Mining processes on the Grid, highlighting the different application types and the benefits and drawbacks in each case. Finally, the methodology is validated through real-world case studies, deployed over the EGEE grid environment. @inproceedings{2009ChrysopoulosECEDSC, | |
Christos Dimou, Fani A. Tzima, Andreas Symeonidis and Pericles Mitkas
"Specifying and Validating the Agent Performance Evaluation Methodology: The Symbiosis Use Case"
IADIS International Conference on Intelligent Systems and Agents, Algarve, Portugal, 2009 Jun
@inproceedings{2009DimouIADIS, | |
Manolis Falelakis, Christos Maramis, Irini Lekka, Pericles Mitkas and Anastasios Delopoulos
"An Ontology for Supporting Clincal Research on Cervical Cancer"
International Conference on Knowledge Engineering and Ontology Development, pp. 103--108, Springer-Verlag, Madeira, Portugal, 2009 Jan
This work presents an ontology for cervical cancer that is positioned in the center of a research system for conducting association studies. The ontology aims at providing a uni?ed ”language” for various heterogeneous medical repositories. To this end, it contains both generic patient-management and domain-speci?c concepts, as well as proper uni?cation rules. The inference scheme adopted is coupled with a procedural programming layer in order to comply with the design requirements. @inproceedings{2009FalelakisICKEOD, | |
Konstantinos M. Karagiannis, Fotis E. Psomopoulos and Pericles A. Mitkas
"Multi Level Clustering of Phylogenetic Profiles"
4th Conference of the Hellenic Society For Computational Biology and Bioinformatics - HSCBB '09, Athens, Greece, 2009 Dec
The prediction of gene function from genome sequences is one of the main issues in Bioinformatics. Most computational approaches are based on the similarity between sequences to infergene function. However, the availability of several fully sequenced genomes has enabled alternative approaches, such as phylogenetic profiles (Pellegriniet al. (1999)). Phylogenetic profiles (pp) are vectors which indicate the presence or absence of a gene in other genomes. The main concept of pp’s is that proteins participating in a common structural complex or metabolic pathway are likely to evolve in a correlated fashion. In this paper, a multi level clustering algorithm of pp’s is presented, which aims to detect inter- and intra-genome gene clusters @inproceedings{2009KaragiannisHSCBB, | |
Pericles A. Mitkas, Anastasios Ntelopoulos, Konstantinos N. Vavliakis, Christos Maramis and Manolis Falelakis andSotiris Diplaris andKoutkias Vasilis andLekka Irini andA. Tantsis andT. Mikos andNikolaos Maglaveras andTheodoros Agorastos
"Pooling data from different sources towards cervical cancer prevention - The ASSIST Project"
8th Scientific Meeting, New Developments in Prevention and Confrontation of Gynecological Cancer, Thessaloniki, Greece, 2009 Jan
@inproceedings{2009MitkasNDPCGC, | |
Vivia Nikolaidou and Pericles A. Mitkas
"A Sequence Mining Method to Predict the Bidding Strategy of Trading Agents"
4th International Workshop on Agents and Data Mining Interaction (ADMI 2009), pp. 139-151, Springer-Verlag, Berlin, Heidelberg, 2009 Jan
In this work, we describe the process used in order to predict the bidding strategy of trading agents. This was done in the context of the Reverse TAC, or CAT, game of the Trading Agent Competition. In this game, a set of trading agents, buyers or sellers, are provided by the server and they trade their goods in one of the markets operated by the competing agents. Better knowledge of the strategy of the trading agents will allow a market maker to adapt its incentives and attract more agents to its own market. Our prediction was based on the time series of the traders\\' past bids, taking into account the variation of each bid compared to its history. The results proved to be of satisfactory accuracy, both in the game\\'s context and when compared to other existing approaches. @inproceedings{2009NikolaidouADMI, | |
John E. Psaroudakis, Fani A. Tzima and Pericles A. Mitkas
"EVADING: An Evolutionary Algorithm with Dynamic Niching for Data Classification"
2009 International Conference on Genetic and Evolutionary Methods (GEM, pp. 59--65, Las Vegas, Nevada, USA, 2009 Jul
Multimodal optimization problems (MMOPs) have been widely studied in many fields of machine learning, including pattern recognition and data classification. Formulating the process of rule induction for the latter task as a MMOP and inspired by corresponding findings in the field of function optimization, our current work proposes an evolutionary algorithm (EVADING) capable of discovering a set of accurate and diverse classification rules. The proposed algorithm uses a dynamic clustering technique as a parallel niching method to maintain rule population diversity and converge to the optimal rules for the attribute-space defined by the target dataset. To demonstrate its applicability and potential, EVADING is applied to a series of real-life classification problems and its prediction accuracy is compared to that of other popular non-evolutionary machine learning techniques. Results are encouraging, since EVADING manages to achieve the best overall average ranking and performs significantly better (at significance level a @inproceedings{2009PsaroudakisGEM, | |
Marina Riga, Fani A. Tzima, Kostas Karatzas and Pericles A. Mitkas
"Development and evaluation of data mining models for air quality prediction in Athens, Greece"
Information Technologies in Environmental Engineering, Proceedings of the 4th International ICSC Symposium, ITEE 2009, pp. 331--344, Springer Berlin Heidelberg, Thessaloniki, Greece, 2009 May
Air pollution is a major problem in the world today, causing undesirable effects on both the environment and human health and, at the same time, stressing the need for effective simulation and forecasting models of atmospheric quality. Targeting this adverse situation, our current work focuses on investigating the potential of data mining algorithms in air pollution modeling and short-term forecasting problems. In this direction, various data mining methods are adopted for the qualitative forecasting of concentration levels of air pollutants or the quantitative prediction of their values (through the development of different classification and regression models respectively) in five locations of the greater Athens area. An additional aim of this work is the systematic assessment of the quality of experimental results, in order to discover the best performing algorithm (or set of algorithms) that can be proved to be significantly different from its rivals. Obtained experimental results are deemed satisfactory in terms of the aforementioned goals of the investigation, as high percentages of correct classifications are achieved in the set of monitoring stations and clear conclusions are drawn, as far as the determination of significantly best performing algorithms is concerned, for the development of air quality (AQ) prediction models. @inproceedings{2009TzimaITEE, |
2008
Conference Papers
Kyriakos C. Chatzidimitriou, Andreas L. Symeonidis and Pericles A. Mitkas
"Data Mining-Driven Analysis and Decomposition in Agent Supply Chain Management Networks"
IEEE/WIC/ACM Workshop on Agents and Data Mining Interaction, pp. 558-561, IEEE Computer Society, Sydney, Australia, 2008 Dec
In complex and dynamic environments where interdependencies cannot monotonously determine causality, data mining techniques may be employed in order to analyze the problem, extract key features and identify pivotal factors. Typical cases of such complexity and dynamicity are supply chain networks, where a number of involved stakeholders struggle towards their own benefit. These stakeholders may be agents with varying degrees of autonomy and intelligence, in a constant effort to establish beneficiary contracts and maximize own revenue. In this paper, we illustrate the benefits of data mining analysis on a well-established agent supply chain management network. We apply data mining techniques, both at a macro and micro level, analyze the results and discuss them in the context of agent performance improvement. @inproceedings{2008ChatzidimitriouADMI, | |
Christos N. Gkekas, Fotis E. Psomopoulos and Pericles A. Mitkas
"Exploiting parallel data mining processing for protein annotation"
Student EUREKA 2008: 2nd Panhellenic Scientific Student Conference, pp. 242-252, Samos, Greece, 2008 Aug
Proteins are large organic compounds consisting of amino acids arranged in a linear chain and joined together by peptide bonds. One of the most important challenges in modern Bioinformatics is the accurate prediction of the functional behavior of proteins. In this paper a novel parallel methodology for automatic protein function annotation is presented. Data mining techniques are employed in order to construct models based on data generated from already annotated protein sequences. The first step of the methodology is to obtain the motifs present in these sequences, which are then provided as input to the data mining algorithms in order to create a model for every term. Experiments conducted using the EGEE Grid environment as a source of multiple CPUs clearly indicate that the methodology is highly efficient and accurate, as the utilization of many processors substantially reduces the execution time. @inproceedings{2008CkekasEURECA, | |
Christos Dimou, Manolis Falelakis, Andreas Symeonidis, Anastasios Delopoulos and Pericles A. Mitkas
"Constructing Optimal Fuzzy Metric Trees for Agent Performance Evaluation"
IEEE/WIC/ACM International Joint Conference on Web Intelligence and Intelligent Agent Technology (WI-IAT\9208), pp. 336--339, IEEE Computer Society, Sydney, Australia, 2008 Dec
The field of multi-agent systems has reached a significant degree of maturity with respect to frameworks, standards and infrastructures. Focus is now shifted to performance evaluation of real-world applications, in order to quantify the practical benefits and drawbacks of agent systems. Our approach extends current work on generic evaluation methodologies for agents by employing fuzzy weighted trees for organizing evaluation-specific concepts/metrics and linguistic terms to intuitively represent and aggregate measurement information. Furthermore, we introduce meta-metrics that measure the validity and complexity of the contribution of each metric in the overall performance evaluation. These are all incorporated for selecting optimal subsets of metrics and designing the evaluation process in compliance with the demands/restrictions of various evaluation setups, thus minimizing intervention by domain experts. The applicability of the proposed methodology is demonstrated through the evaluation of a real-world test case. @inproceedings{2008DimouIAT, | |
Christos Dimou, Kyriakos C. Chatzidimitriou, Andreas Symeonidis and Pericles A. Mitkas
"Creating and Reusing Metric Graphs for Evaluating Agent Performance in the Supply Chain Management Domain"
First Workshop on Knowledge Reuse (KREUSE, Beijing (China), 2008 May
The overwhelming demand for efficient agent performance in Supply Chain Management systems, as exemplified by numerous international competitions, raises the issue of defining and using generalized methods for performance evaluation. Up until now, most researchers test their findings in an ad-hoc manner, often having to re-invent existing evaluation-specific knowledge. In this position paper, we tackle the key issue of defining and using metrics within the context of evaluating agent performance in the SCM domain. We propose the Metrics Representation Graph, a structure that organizes performance metrics in hierarchical manner, and perform a preliminary assessment by instantiating an MRG for the TAC SCM competition, one of the most demanding SCM competitions currently established. We envision the automated generation of the MRG, as well as appropriate contribution from the TAC community towards the finalization of the MRG, so that it will be readily available for future performance evaluations. @inproceedings{2008DimouKREUSE, | |
Christos Dimou, Andreas L. Symeonidis and Pericles A. Mitkas
"Data Mining and Agent Technology: a fruitful symbiosis"
Soft Computing for Knowledge Discovery and Data Mining, pp. 327-362, Springer US, Clermont-Ferrand, France, 2008 Jan
Multi-agent systems (MAS) have grown quite popular in a wide spec- trum of applications where argumentation, communication, scaling and adaptability are requested. And though the need for well-established engineering approaches for building and evaluating such intelligent systems has emerged, currently no widely accepted methodology exists, mainly due to lack of consensus on relevant defini- tions and scope of applicability. Even existing well-tested evaluation methodologies applied in traditional software engineering, prove inadequate to address the unpre- dictable emerging factors of the behavior of intelligent components. The following chapter aims to present such a unified and integrated methodology for a specific cat- egory of MAS. It takes all constraints and issues into account and denotes the way knowledge extracted with the use of Data mining (DM) techniques can be used for the formulation initially, and the improvement, in the long run, of agent reasoning and MAS performance. The coupling of DM and Agent Technology (AT) principles, proposed within the context of this chapter is therefore expected to provide to the reader an efficient gateway for developing and evaluating highly reconfigurable soft- ware approaches that incorporate domain knowledge and provide sophisticated De- cision Making capabilities. The main objectives of this chapter could be summarized into the following: a) introduce Agent Technology (AT) as a successful paradigm for building Data Mining (DM)-enriched applications, b) provide a methodology for (re)evaluating the performance of such DM-enriched Multi-Agent Systems (MAS), c) Introduce Agent Academy II, an Agent-Oriented Software Engineering framework for building MAS that incorporate knowledge model extracted by the use of (classi- cal and novel) DM techniques and d) denote the benefits of the proposed approach through a real-world demonstrator. This chapter provides a link between DM and AT and explains how these technologies can efficiently cooperate with each other. The exploitation of useful knowledge extracted by the use of DM may consider- ably improve agent infrastructures, while also increasing reusability and minimizing customization costs. The synergy between DM and AT is ultimately expected to provide MAS with higher levels of autonomy, adaptability and accuracy and, hence, intelligence. @inproceedings{2008DimouSCKDDM, | |
Christos N. Gkekas, Fotis E. Psomopoulos and Pericles A. Mitkas
"A Parallel Data Mining Application for Gene Ontology Term Prediction"
3rd EGEE User Forum, Clermont-Ferrand, France, 2008 Feb
One of the most important challenges in modern bioinformatics is the accurate prediction of the functional behaviour of proteins. The strong correlation that exists between the properties of a protein and its motif sequence makes such a prediction possible. In this paper a novel parallel methodology for protein function prediction will be presented. Data mining techniques are employed in order to construct a model for each Gene Ontology term, based on data generated from already annotated protein sequences. In order to predict the annotation of an unknown protein, its motif sequence is run through each GO term model, producing similarity scores for every term. Although it has been experimentally proven that this process is efficient, it unfortunately requires heavy processor resources. In order to address this issue, a parallel application has been implemented and tested using the EGEE Grid infrastructure. @inproceedings{2008GkekasEGEEForum, | |
Christos N. Gkekas, Fotis E. Psomopoulos and Pericles A. Mitkas
"A Parallel Data Mining Methodology for Protein Function Prediction Utilizing Finite State Automata"
2nd Electrical and Computer Engineering Student Conference, Athens, Greece, 2008 Apr
One of the most important challenges in modern bioinformatics is the accurate prediction of the functional behaviour of proteins. The strong correlation that exists between the properties of a protein and its motif sequence makes such a prediction possible. In this paper a novel parallel methodology for protein function prediction will be presented. Data mining techniques are employed in order to construct a model for each Gene Ontology term, based on data generated from already annotated protein sequences. In order to predict the annotation of an unknown protein, its motif sequence is run through each GO term model, producing similarity scores for every term. Although it has been experimentally proven that this process is efficient, it unfortunately requires heavy processor resources. In order to address this issue, a parallel application has been implemented and tested using the EGEE Grid infrastructure. @inproceedings{2008GkekasSFHMMY, | |
Georgios Karagiannis, Konstantinos N. Vavliakis, Stella Markantonatou, Sister Daniilia, Sophia Sotiropoulou, Maria Alexopoulou, Olga Yanoutsou, Klimis Dalianis and Thodoros Kavalieros
"EIKONOGNOSIA An Integrated System for Advanced Retrieval of Scientific Data and Metadata of Byzantine Artworks Using Semantic Web Technologies"
Annual Conference of CIDOC, pp. 558-561, IEEE Computer Society, Athens, Greece, 2008 Sep
The documentation and analysis of Byzantine Art is an important component of the overall effort to maintain cultural heritage and contributes to learning and comprehending ones history traversal path. Efficient publishing of the multi-dimensional and multifaceted information that is necessary for the complete documentation of artworks should draw on a good organization of the data. Eikonognosia is a research project funded by the Greek General Secretariat of Research and Technology (GSRT) that aims to efficiently organize and publish detailed information about icons in the World Wide Web. Information derived from the analysis conducted in the Art Diagnosis Center of Ormylia Foundation is taken as a case study. Eikonognosia provides the means for organising detailed and multidimensional information about Byzantine icons in a way that is compatible to international standards (CIDOC-CRM - ISO 21127:2006) and allows for an easy retrieval of data with advanced semantic web technologies. The ultimate goal for Eikonognosia is to foster the cultural heritage community by providing an integrated framework that helps to facilitate organization, retrieval and presentation of data from the cultural heritage domain. @inproceedings{2008KaragiannisCIDOC, | |
Kostas Karatzas, Anastasios S Bassoukos, Dimitris Voukantsis, Fani A. Tzima, Kostas Nikolaou and Stavros Karathanasis
"ICT technologies and computational intelligence methods for the creation of an early warning air pollution information system"
22nd Conference on Environmental Informatics and Industrial Ecology, 2008 Sep
Contemporary air quality management calls for effective, and in advance, AQ information dissemination. Such dissemination requires for communication that should not be based solely on written or oral language forms, but should make use of graphical, symbolical and multimedia language communication schemes, via available communication channels. Previous experiences and published research results indicate that the content of environmental information systems should include both realtime information and forecasting for key parameters of interest, like the maximum concentration values of air pollutants. The latter are difficult to achieve, as air quality forecasting requires both do main expertise and modelling skills for the complicated phenomenon of atmospheric pollution. One of the ways to address this need and to extract useful knowledge for better forecasting and understanding of air pollution problems, is the application of Computational Intelligence (CI) methodsand tools. The present paper discusses the creation of an environmental information portal for the dissemination of air quality information and warnings, for the city of Thessaloniki, Greece. The system is developed with the aid of state-of-the art, web-based technologies, including modular, on the fly software integration to operating applications, and implements CI for the forecasting of parameters of interest. In addition, observation data are made accessible to the public via an internet-based, graphics environment that deploys open source geographic information services. @inproceedings{2008KaratzasCEIIE, | |
Pericles A. Mitkas
"Training Intelligent Agents and Evaluating Their Performance"
International Workshop on Agents and Data Mining Interaction (ADMI), pp. 336--339, IEEE Computer Society, Sydney,Australia, 2008 Dec
The field of multi-agent systems has reached a significant degree of maturity with respect to frameworks, standards and infrastructures. Focus is now shifted to performance evaluation of real-world applications, in order to quantify the practical benefits and drawbacks of agent systems. Our approach extends current work on generic evaluation methodologies for agents by employing fuzzy weighted trees for organizing evaluation-specific concepts/metrics and linguistic terms to intuitively represent and aggregate measurement information. Furthermore, we introduce meta-metrics that measure the validity and complexity of the contribution of each metric in the overall performance evaluation. These are all incorporated for selecting optimal subsets of metrics and designing the evaluation process in compliance with the demands/restrictions of various evaluation setups, thus minimizing intervention by domain experts. The applicability of the proposed methodology is demonstrated through the evaluation of a real-world test case. @inproceedings{2008MitkasADMI, | |
Pericles A. Mitkas, Christos Maramis, Anastastios N. Delopoulos, Andreas Symeonidis, Sotiris Diplaris, Manolis Falelakis, Fotis E. Psomopoulos, Alex andros Batzios, Nikolaos Maglaveras, Irini Lekka, Vasilis Koutkias, Theodoros Agorastos, T. Mikos and A. Tatsis
"ASSIST: Employing Inference and Semantic Technologies to Facilitate Association Studies on Cervical Cancer"
6th European Symposium on Biomedical Engineering, Chania, Greece, 2008 Jun
Despite the proved close connection of cervical cancer with the human papillomavirus (HPV), intensive ongoing research investigates the role of specific genetic and environmental factors in determining HPV persistence and subsequent progression of the disease. To this end, genetic association studies constitute a significant scientific approach that may lead to a more comprehensive insight on the origin of complex diseases. Nevertheless, association studies are most of the times inconclusive, since the datasets employed are small, usually incomplete and of poor quality. The main goal of ASSIST is to aid research in the field of cervical cancer providing larger high quality datasets, via a software system that virtually unifies multiple heterogeneous medical records, located in various sites. Furthermore, the system is being designed in a generic manner, with provision for future extensions to include other types of cancer or even different medical fields. Within the context of ASSIST, innovative techniques have been elaborated for the semantic modelling and fuzzy inferencing on medical knowledge aiming at meaningful data unification: (i) The ASSIST core ontology (being the first ontology ever modelling cervical cancer) permits semantically equivalent but differently coded data to be mapped to a common language. (ii) The ASSIST inference engine maps medical entities to syntactic values that are understood by legacy medical systems, supporting the processes of hypotheses testing and association studies, and at the same time calculating the severity index of each patient record. These modules constitute the ASSIST Core and are accompanied by two other important subsystems: (1) The Interfacing to Medical Archives subsystem maps the information contained in each legacy medical archive to corresponding entities as defined in the knowledge model of ASSIST. These patient data are generated by an advanced anonymisation tool also developed within the context of the project. (2) The User Interface enables transparent and advanced access to the data repositories incorporated in ASSIST by offering query expression as well as patient data and statistical results visualisation to the ASSIST end-users. We also have to point out that the system is easily extendable virtually to any medical domain, as the core ontology was designed with this in mind and all subsystems are ontology-aware i.e., adaptable to any ontology changes/additions. Using ASSIST, a medical researcher can have seamless access to medical records of participating sites and, through a particularly handy computing environment, collect data records satisfying his criteria. Moreover he can define cases and controls, select records adjusting their validity and use the most popular statistical tools for drawing conclusions. The logical unification of medical records of participating sites, including clinical and genetic data, to a common knowledge base is expected to increase the effectiveness of research in the field of cervical cancer as it permits the creation of on-demand study groups as well as the recycling of data used in previous studies. @inproceedings{2008MitkasEsbmeAssist, | |
Pericles A. Mitkas, Vassilis Koutkias, Andreas Symeonidis, Manolis Falelakis, Christos Diou, Irini Lekka, Anastasios T. Delopoulos, Theodoros Agorastos and Nicos Maglaveras
"Association Studies on Cervical Cancer Facilitated by Inference and Semantic Technologes: The ASSIST Approach"
MIE, Goteborg, Sweden, 2008 May
Cervical cancer (CxCa) is currently the second leading cause of cancer-related deaths, for women between 20 and 39 years old. As infection by the human papillomavirus (HPV) is considered as the central risk factor for CxCa, current research focuses on the role of specific genetic and environmental factors in determining HPV persistence and subsequent progression of the disease. ASSIST is an EU-funded research project that aims to facilitate the design and execution of genetic association studies on CxCa in a systematic way by adopting inference and semantic technologies. Toward this goal, ASSIST provides the means for seamless integration and virtual unification of distributed and heterogeneous CxCa data repositories, and the underlying mechanisms to undertake the entire process of expressing and statistically evaluating medical hypotheses based on the collected data in order to generate medically important associations. The ultimate goal for ASSIST is to foster the biomedical research community by providing an open, integrated and collaborative framework to facilitate genetic association studies. @conference{2008MitkasMIE, | |
Ioanna K. Mprouza, Fotis E. Psomopoulos and Pericles A. Mitkas
"AMoS: Agent-based Molecular Simulations"
Student EUREKA 2008: 2nd Panhellenic Scientific Student Conference, pp. 175-186, Samos, Greece, 2008 Aug
Molecular dynamics (MD) is a form of computer simulation wherein atoms and molecules are allowed to interact for a period of time under known laws of physics, giving a view of the motion of the atoms. Usually the number of particles involved in a simulation is so large, that the properties of the system in question are virtually impossible to compute analytically. MD circumvents this problem by employing numerical approaches. Utilizing theories and concepts from mathematics, physics and chemistry and employing algorithms from computer science and information theory, MD is a clear example of a multidisciplinary method. In this paper a new framework for MD simulations is presented, which utilizes software agents as particle representations and an empirical potential function as the means of interaction. The framework is applied on protein structural data (PDB files), using an implicit solvent environment and a time step of 5 femto-seconds (5×10−15 sec). The goal of the simulation is to provide another view to the study of emergent behaviours and trends in the movement of the agent-particles in the protein complex. This information can then be used to construct an abstract model of the rules that govern the motion of the particles. @inproceedings{2008MprouzaEURECA, | |
Fotis E. Psomopoulos and Pericles A. Mitkas
"Sizing Up: Bioinformatics in a Grid Context"
3rd Conference of the Hellenic Society For Computational Biology and Bioinformatics - HSCBB, pp. 558-561, IEEE Computer Society, Thessaloniki, Greece, 2008 Oct
A Frid environmeent can be viewed sa a virtual computing architecture that provides the ability to perform higher thoughput computing by taking advantage of many computer geographically distributed and connected by a network. Bioinformatics applications stand to gain in such environment both in regards of cimputational resources available, but in reliability and efficiency as well. There are several approaches in literature which present the use of Grid resources in bioinformatics. Nevertheless, scientific progress is hindered by the fact that each researcher operates in relative isolation, regarding datasets and efforts, since there is no universally accepted methodology for performing bioinformatics tasks in Grid. Given the complexity of both the data and the algorithms invilvde in the majorityof cases, a case study on protein classification utilizing the Frid infrastructure, may be the first step in presenting a unifying methodology for bioinformatics in a Grind context. @inproceedings{2008PsomopoulosHSCBB, | |
Fotis E. Psomopoulos, Pericles A. Mitkas, Christos S. Krinas and Ioannis N. Demetropoulos
"G-MolKnot: A grid enabled systematic algorithm to produce open molecular knots"
1st HellasGrid User Forum, pp. 327-362, Springer US, Athens, Greece, 2008 Jan
Multi-agent systems (MAS) have grown quite popular in a wide spec- trum of applications where argumentation, communication, scaling and adaptability are requested. And though the need for well-established engineering approaches for building and evaluating such intelligent systems has emerged, currently no widely accepted methodology exists, mainly due to lack of consensus on relevant defini- tions and scope of applicability. Even existing well-tested evaluation methodologies applied in traditional software engineering, prove inadequate to address the unpre- dictable emerging factors of the behavior of intelligent components. The following chapter aims to present such a unified and integrated methodology for a specific cat- egory of MAS. It takes all constraints and issues into account and denotes the way knowledge extracted with the use of Data mining (DM) techniques can be used for the formulation initially, and the improvement, in the long run, of agent reasoning and MAS performance. The coupling of DM and Agent Technology (AT) principles, proposed within the context of this chapter is therefore expected to provide to the reader an efficient gateway for developing and evaluating highly reconfigurable soft- ware approaches that incorporate domain knowledge and provide sophisticated De- cision Making capabilities. The main objectives of this chapter could be summarized into the following: a) introduce Agent Technology (AT) as a successful paradigm for building Data Mining (DM)-enriched applications, b) provide a methodology for (re)evaluating the performance of such DM-enriched Multi-Agent Systems (MAS), c) Introduce Agent Academy II, an Agent-Oriented Software Engineering framework for building MAS that incorporate knowledge model extracted by the use of (classi- cal and novel) DM techniques and d) denote the benefits of the proposed approach through a real-world demonstrator. This chapter provides a link between DM and AT and explains how these technologies can efficiently cooperate with each other. The exploitation of useful knowledge extracted by the use of DM may consider- ably improve agent infrastructures, while also increasing reusability and minimizing customization costs. The synergy between DM and AT is ultimately expected to provide MAS with higher levels of autonomy, adaptability and accuracy and, hence, intelligence. @inproceedings{2008PsomopoulosHUF, | |
Fani A. Tzima and Pericles A. Mitkas
"ZCS Revisited: Zeroth-level Classifier Systems for Data Mining"
2008 IEEE International Conference on Data Mining Workshops, pp. 700--709, IEEE Computer Society, Washington, DC, 2008 Dec
Learning classifier systems (LCS) are machine learning systems designed to work for both multi-step and singlestep decision tasks. The latter case presents an interesting, though not widely studied, challenge for such algorithms, especially when they are applied to real-world data mining problems. The present investigation departs from the popular approach of applying accuracy-based LCS to data mining problems and aims to uncover the potential of strengthbased LCS in such tasks. In this direction, ZCS-DM, a Zeroth-level Classifier System for data mining, is applied to a series of real world classification problems and its performance is compared to that of other state-of-the-art machine learning techniques (C4.5, HIDER and XCS). Results are encouraging, since with only a modest parameter exploration phase, ZCS-DM manages to outperform its rival algorithms in eleven out of the twelve benchmark datasets used in this study. We conclude this work by identifying future research directions. @inproceedings{2008TzimaICDMW, | |
Konstantinos N. Vavliakis, Georgios Th. Karagiannis and Sophia Sotiropoulou
"The AKMON Project: Semantic Web in Byzantine Iconography"
Paving the way to a semantic web for cultural heritage, Workshop held in conjunction with Vast 2008 Conference, pp. 327-362, Springer US, Braga, Portugal, 2008 Jan
Multi-agent systems (MAS) have grown quite popular in a wide spec- trum of applications where argumentation, communication, scaling and adaptability are requested. And though the need for well-established engineering approaches for building and evaluating such intelligent systems has emerged, currently no widely accepted methodology exists, mainly due to lack of consensus on relevant defini- tions and scope of applicability. Even existing well-tested evaluation methodologies applied in traditional software engineering, prove inadequate to address the unpre- dictable emerging factors of the behavior of intelligent components. The following chapter aims to present such a unified and integrated methodology for a specific cat- egory of MAS. It takes all constraints and issues into account and denotes the way knowledge extracted with the use of Data mining (DM) techniques can be used for the formulation initially, and the improvement, in the long run, of agent reasoning and MAS performance. The coupling of DM and Agent Technology (AT) principles, proposed within the context of this chapter is therefore expected to provide to the reader an efficient gateway for developing and evaluating highly reconfigurable soft- ware approaches that incorporate domain knowledge and provide sophisticated De- cision Making capabilities. The main objectives of this chapter could be summarized into the following: a) introduce Agent Technology (AT) as a successful paradigm for building Data Mining (DM)-enriched applications, b) provide a methodology for (re)evaluating the performance of such DM-enriched Multi-Agent Systems (MAS), c) Introduce Agent Academy II, an Agent-Oriented Software Engineering framework for building MAS that incorporate knowledge model extracted by the use of (classi- cal and novel) DM techniques and d) denote the benefits of the proposed approach through a real-world demonstrator. This chapter provides a link between DM and AT and explains how these technologies can efficiently cooperate with each other. The exploitation of useful knowledge extracted by the use of DM may consider- ably improve agent infrastructures, while also increasing reusability and minimizing customization costs. The synergy between DM and AT is ultimately expected to provide MAS with higher levels of autonomy, adaptability and accuracy and, hence, intelligence. @inproceedings{2008VavliakisVAST, | |
Theodoros Agorastos, Pericles A. Mitkas, Manolis Falelakis, Fotis E. Psomopoulos, Anastasios N. Delopoulos, Andreas Symeonidis, Sotiris Diplaris, Christos Maramis, Alexandros Batzios, Irini Lekka, Vasilis Koutkias, Themistoklis Mikos, A. Tatsis and Nikolaos Maglaveras
"Large Scale Association Studies Using Unified Data for Cervical Cancer and beyond: The ASSIST Project"
World Cancer Congress, Geneva, Switzerland, 2008 Aug
Despite the proved close connection of cervical cancer with the human papillomavirus (HPV), intensive ongoing research investigates the role of specific genetic and environmental factors in determining HPV persistence and subsequent progression of the disease. To this end, genetic association studies constitute a significant scientific approach that may lead to a more comprehensive insight on the origin of complex diseases. Nevertheless, association studies are most of the times inconclusive, since the datasets employed are small, usually incomplete and of poor quality. The main goal of ASSIST is to aid research in the field of cervical cancer providing larger high quality datasets, via a software system that virtually unifies multiple heterogeneous medical records, located in various sites. Furthermore, the system is being designed in a generic manner, with provision for future extensions to include other types of cancer or even different medical fields. Within the context of ASSIST, innovative techniques have been elaborated for the semantic modelling and fuzzy inferencing on medical knowledge aiming at meaningful data unification: (i) The ASSIST core ontology (being the first ontology ever modelling cervical cancer) permits semantically equivalent but differently coded data to be mapped to a common language. (ii) The ASSIST inference engine maps medical entities to syntactic values that are understood by legacy medical systems, supporting the processes of hypotheses testing and association studies, and at the same time calculating the severity index of each patient record. These modules constitute the ASSIST Core and are accompanied by two other important subsystems: (1) The Interfacing to Medical Archives subsystem maps the information contained in each legacy medical archive to corresponding entities as defined in the knowledge model of ASSIST. These patient data are generated by an advanced anonymisation tool also developed within the context of the project. (2) The User Interface enables transparent and advanced access to the data repositories incorporated in ASSIST by offering query expression as well as patient data and statistical results visualisation to the ASSIST end-users. We also have to point out that the system is easily extendable virtually to any medical domain, as the core ontology was designed with this in mind and all subsystems are ontology-aware i.e., adaptable to any ontology changes/additions. Using ASSIST, a medical researcher can have seamless access to medical records of participating sites and, through a particularly handy computing environment, collect data records satisfying his criteria. Moreover he can define cases and controls, select records adjusting their validity and use the most popular statistical tools for drawing conclusions. The logical unification of medical records of participating sites, including clinical and genetic data, to a common knowledge base is expected to increase the effectiveness of research in the field of cervical cancer as it permits the creation of on-demand study groups as well as the recycling of data used in previous studies. @inproceedings{WCCAssist, |
2007
Conference Papers
Chrysa Collyda, Sotiris Diplaris, Pericles A. Mitkas, Nicos Maglaveras and Costas Pappas
"Profile Fuzzy Hidden Markov Models for Phylogenetic Analysis and Protein Classification"
5th Annual Rocky Mountain Bioinformatics Conference, pp. 327-362, Springer US, Aspen/Snowmass, CO, USA, 2007 Nov
Multi-agent systems (MAS) have grown quite popular in a wide spec- trum of applications where argumentation, communication, scaling and adaptability are requested. And though the need for well-established engineering approaches for building and evaluating such intelligent systems has emerged, currently no widely accepted methodology exists, mainly due to lack of consensus on relevant defini- tions and scope of applicability. Even existing well-tested evaluation methodologies applied in traditional software engineering, prove inadequate to address the unpre- dictable emerging factors of the behavior of intelligent components. The following chapter aims to present such a unified and integrated methodology for a specific cat- egory of MAS. It takes all constraints and issues into account and denotes the way knowledge extracted with the use of Data mining (DM) techniques can be used for the formulation initially, and the improvement, in the long run, of agent reasoning and MAS performance. The coupling of DM and Agent Technology (AT) principles, proposed within the context of this chapter is therefore expected to provide to the reader an efficient gateway for developing and evaluating highly reconfigurable soft- ware approaches that incorporate domain knowledge and provide sophisticated De- cision Making capabilities. The main objectives of this chapter could be summarized into the following: a) introduce Agent Technology (AT) as a successful paradigm for building Data Mining (DM)-enriched applications, b) provide a methodology for (re)evaluating the performance of such DM-enriched Multi-Agent Systems (MAS), c) Introduce Agent Academy II, an Agent-Oriented Software Engineering framework for building MAS that incorporate knowledge model extracted by the use of (classi- cal and novel) DM techniques and d) denote the benefits of the proposed approach through a real-world demonstrator. This chapter provides a link between DM and AT and explains how these technologies can efficiently cooperate with each other. The exploitation of useful knowledge extracted by the use of DM may consider- ably improve agent infrastructures, while also increasing reusability and minimizing customization costs. The synergy between DM and AT is ultimately expected to provide MAS with higher levels of autonomy, adaptability and accuracy and, hence, intelligence. @inproceedings{2007CollydaARMBC, | |
Christos Dimou, Andreas L. Symeonidis and Pericles A. Mitkas
"Evaluating Knowledge Intensive Multi-Agent Systems"
Autonomous Intelligent Systems: Multi-Agents and Data Mining (AIS-ADM 2007), pp. 74-87, Springer Berlin / Heidelberg, St. Petersburg, Russia, 2007 Jun
As modern applications tend to stretch between large, evergrowing datasets and increasing demand for meaningful content at the user end, more elaborate and sophisticated knowledge extraction technologies are needed. Towards this direction, the inherently contradicting technologies of deductive software agents and inductive data mining have been integrated, in order to address knowledge intensive problems. However, there exists no generalized evaluation methodology for assessing the efficiency of such applications. On the one hand, existing data mining evaluation methods focus only on algorithmic precision, ignoring overall system performance issues. On the other hand, existing systems evaluation techniques are insufficient, as the emergent intelligent behavior of agents introduce unpredictable factors of performance. In this paper, we present a generalized methodology for performance evaluation of intelligent agents that employ knowledge models produced through data mining. The proposed methodology consists of concise steps for selecting appropriate metrics, defining measurement methodologies and aggregating the measured performance indicators into thorough system characterizations. The paper concludes with a demonstration of the proposed methodology to a real world application, in the Supply Chain Management domain. @inproceedings{2007DimouAIS, | |
Christos Dimou, Andreas L. Symeonidis and Pericles A. Mitkas
"Towards a Generic Methodology for Evaluating MAS Performance"
IEEE International Conference on Integration of Knowledge Intensive Multi-Agents Systems - KIMAS\9207, pp. 174--179, Springer Berlin / Heidelberg, Waltham, MA, USA, 2007 Apr
As Agent Technology (AT) becomes a well-established engineering field of computing, the need for generalized, standardized methodologies for agent evaluation is imperative. Despite the plethora of available development tools and theories that researchers in agent computing have access to, there is a remarkable lack of general metrics, tools, benchmarks and experimental methods for formal validation and comparison of existing or newly developed systems. It is argued that AT has reached a certain degree of maturity, and it is therefore feasible to move from ad-hoc, domain- specific evaluation methods to standardized, repeatable and easily verifiable procedures. In this paper, we outline a first attempt towards a generic evaluation methodology for MAS performance. Instead of following the research path towards defining more powerful mathematical description tools for determining intelligence and performance metrics, we adopt an engineering point of view to the problem of deploying a methodology that is both implementation and domain independent. The proposed methodology consists of a concise set of steps, novel theoretical representation tools and appropriate software tools that assist evaluators in selecting the appropriate metrics, undertaking measurement and aggregation techniques for the system at hand. @inproceedings{2007DimouKIMAS, | |
Christos Dimou, Andreas L. Symeonidis and Pericles A. Mitkas
"An agent structure for evaluating micro-level MAS performance"
7th Workshop on Performance Metrics for Intelligent Systems - PerMIS-07, pp. 243--250, IEEE Computer Society, Gaithersburg, MD, 2007 Aug
Although the need for well-established engineering approaches in Intelligent Systems (IS) performance evaluation is urging, currently no widely accepted methodology exists, mainly due to lackof consensus on relevant definitions and scope of applicability, multi-disciplinary issues and immaturity of the field of IS. Even existing well-tested evaluation methodologies applied in other domains, such as (traditional) software engineering, prove inadequate to address the unpredictable emerging factors of the behavior of intelligent components. In this paper, we present a generic methodology and associated tools for evaluating the performance of IS, by exploiting the software agent paradigm as a representative modeling concept for intelligent systems. Based on the assessment of observable behavior of agents or multi-agent systems, the proposed methodology provides a concise set of guidelines and representation tools for evaluators to use. The methodology comprises three main tasks, namely met ricsselection, monitoring agent activities for appropriate me asurements, and aggregation of the conducted measurements. Coupled to this methodology is the Evaluator Agent Framework, which aims at the automation of most of the provided steps of the methodology, by providing Graphical User Interfaces for metrics organization and results presentation, as well as a code generating module that produces a skeleton of a monitoring agent. Once this agent is completed with domain-specific code, it is appended to the runtime of a multi-agent system and collects information from observable events and messages. Both the evaluation methodology and the automation framework are tested and demonstrated in Symbiosis, a MAS simulation environment for competing groups of autonomous entities. @inproceedings{2007DimouPERMIS, | |
Sotiris Diplaris, G. Papachristoudis and Pericles A. Mitkas
"SoFoCles: Feature Filtering for Microarray Classification Based on Gene Ontology"
Hellenic Bioinformatics and Medical Informatics Meeting, pp. 279--282, IEEE Computer Society, Athens, Greece, 2007 Oct
Semantic annotation and querying is currently applied on a number of versatile disciplines, providing the addedvalue of such an approach and, consequently the need for more elaborate \\\\96 either case-specific or generic \\\\96 tools. In this context, we have developed Eikonomia: an integrated semantically-aware tool for the description and retrieval of Byzantine Artwork Information. Following the needs of the ORMYLIA Art Diagnosis Center for adding semantics to their legacy data, an ontology describing Byzantine artwork based on CIDOCCRM, along with the interfaces for synchronization to and from the existing RDBMS have been implemented. This ontology has been linked to a reasoning tool, while a dynamic interface for the automated creation of semantic queries in SPARQL was developed. Finally, all the appropriate interfaces were instantiated, in order to allow easy ontology manipulation, query results projection and restrictions creation. @inproceedings{2007DiplarisHBMIM, | |
Christos N. Gkekas, Fotis E. Psomopoulos and Pericles A. Mitkas
"Modeling Gene Ontology Terms using Finite State Automata"
Hellenic Bioinformatics and Medical Informatics Meeting, pp. 279--282, IEEE Computer Society, Biomedical Research Foundation, Academy of Athens, Greece, 2007 Oct
Semantic annotation and querying is currently applied on a number of versatile disciplines, providing the addedvalue of such an approach and, consequently the need for more elaborate \\\\96 either case-specific or generic \\\\96 tools. In this context, we have developed Eikonomia: an integrated semantically-aware tool for the description and retrieval of Byzantine Artwork Information. Following the needs of the ORMYLIA Art Diagnosis Center for adding semantics to their legacy data, an ontology describing Byzantine artwork based on CIDOCCRM, along with the interfaces for synchronization to and from the existing RDBMS have been implemented. This ontology has been linked to a reasoning tool, while a dynamic interface for the automated creation of semantic queries in SPARQL was developed. Finally, all the appropriate interfaces were instantiated, in order to allow easy ontology manipulation, query results projection and restrictions creation. @inproceedings{2007GkekasBioacademy, | |
Ioanna K. Mprouza, Fotis E. Psomopoulos and Pericles A. Mitkas
"Simulating molecular dynamics through intelligent software agents"
Hellenic Bioinformatics and Medical Informatics Meeting, pp. 279--282, IEEE Computer Society, Biomedical Research Foundation, Academy of Athens, Greece, 2007 Oct
Semantic annotation and querying is currently applied on a number of versatile disciplines, providing the addedvalue of such an approach and, consequently the need for more elaborate \\\\96 either case-specific or generic \\\\96 tools. In this context, we have developed Eikonomia: an integrated semantically-aware tool for the description and retrieval of Byzantine Artwork Information. Following the needs of the ORMYLIA Art Diagnosis Center for adding semantics to their legacy data, an ontology describing Byzantine artwork based on CIDOCCRM, along with the interfaces for synchronization to and from the existing RDBMS have been implemented. This ontology has been linked to a reasoning tool, while a dynamic interface for the automated creation of semantic queries in SPARQL was developed. Finally, all the appropriate interfaces were instantiated, in order to allow easy ontology manipulation, query results projection and restrictions creation. @inproceedings{2007MprouzaBioacademy, | |
P. Tsimpos, Sotiris Diplaris, Pericles A. Mitkas and Georgios Banos
"Mendelian Samples Mining and Cluster Monitoring for National Genetic Evaluations with AGELI"
Interbull Annual Meeting, pp. 73-77, Dublin, Ireland, 2007 Aug
We present an innovative approach for pre-processing, analysis, alarm issuing and presentation of national genetic evaluation data with AGELI using Mendelian sampling mining and clustering techniques. AGELI (Eleftherohorinou et al.,2005) is a software platform that integrates the whole data mining procedure in order to produce a qualitative description of national genetic evaluation results, concerning three milk yield traits. Quality assurance constitutes a critical issue in the range of services provided by Interbull. Although the standard method appears sufficiently functional (Klei et al.,2002), during the last years there has been progress concerning an alternative validation method of genetic evaluation results using data mining (Banoset al.,2003; Diplaris et al.,2004), potentially leading to inference on data quality. This methodology was incorporated in AGELI in order to assess and assure data quality. The whole idea waImport your BibTex here!! :Ds to exploit decision trees and apply a goodness of fit test to individual tree nodes and an F-test to corresponding nodes from consecutive evaluation runs, aiming at discovering possible abnormalities in bull proof distributions. In a previous report (Banos et al.,2003) predictions led to associations, which were qualitatively compared to actual proofs, and existing discrepancies were confirmed using a data set with known errors. In this report we present AGELI’s novel methods of performing data mining by using a series of decision tree and clustering algorithms. Different decision tree models can now be created in order to assess data quality by evaluating data with various criteria. To further assess data quality, a novel technique for cluster monitoring is implemented in AGELI. It is possible to form clusters of bulls and perform unsupervised monitoring on them over the entire period of genetic evaluation runs. Finally, analyses were conducted using bull Mendelian sampling over the whole dataset. @inproceedings{2007TsimposIAM, | |
Fani A. Tzima, Kostas D. Karatzas, Pericles A. Mitkas and Stavros Karathanasis
"Using data-mining techniques for PM10 forecasting in the metropolitan area of Thessaloniki, Greece"
IJCNN 2007 International - Joint Conference on Neural Netwroks, pp. 2752--2757, Orlando, Florida, 2007 Aug
Knowledge extraction and acute forecasting are among the most challenging issues concerning the use of computational intelligence (CI) methods in real world applications. Both aspects are essential in cases where decision making is required, especially in domains directly related to the quality of life, like the quality of the atmospheric environment. In the present paper we emphasize on short term Air Quality (AQ) forecasting as a key constituent of every AQ management system, and we apply various CI methods and tools for assessing PM10 concentration values. We report our experimental strategy and preliminary results that reveal interesting interrelations between AQ and various city operations, while performing satisfactory in predicting concentration values. @inproceedings{2007TzimaIJCNN, | |
Fani A. Tzima, Andreas L. Symeonidis and Pericles. A. Mitkas
"Symbiosis: using predator-prey games as a test bed for studying competitive coevolution"
IEEE KIMAS conference, pp. 115-120, Springer Berlin / Heidelberg, Waltham, Massachusetts, 2007 Apr
The animat approach constitutes an intriguing attempt to study and comprehend the behavior of adaptive, learning entities in complex environments. Further inspired by the notions of co-evolution and evolutionary arms races, we have developed Symbiosis, a virtual ecosystem that hosts two self-organizing, combating species \\\\96 preys and predators. All animats live and evolve in this shared environment, they are self-maintaining and engage in a series of vital activities nutrition, growth, communication with the ultimate goals of survival and reproduction. The main objective of Symbiosis is to study the behavior of ecosystem members, especially in terms of the emergent learning mechanisms and the effect of co-evolution on the evolved behavioral strategies. In this direction, several indicators are used to assess individual behavior, with the overall effectiveness metric depending strongly on the animats net energy gain and reproduction rate. Several experiments have been conducted with the developed simulator under various environmental conditions. Overall experimental results support our original hypothesis that co-evolution is a driving factor in the animat learning procedure. @inproceedings{2007TzimaKIMAS, | |
Fani A.Tzima, Ioannis N. Athanasiadis and Pericles A. Mitkas
"Agent-based modelling and simulation in the irrigation management sector: applications and potential"
Options Mediterraneennes, Series B: Studies and Research, Proceedings of the WASAMED International Conference, pp. 273--286, 2007 Feb
In the field of sustainable development, the management of common-pool resources is an issue of major importance. Several models that attempt to address the problem can be found in the literature, especially in the case of irrigation management. In fact, the latter task represents a great challenge for researchers and decision makers, as it has to cope with various water-related activities and conflicting user perspectives within a specified geographic area. Simulation models, and particularly Agent-Based Modelling and Simulation (ABMS), can facilitate overcoming these limitations: their inherent ability of integrating ecological and socio-economic dimensions, allows their effective use as tools for evaluating the possible effects of different management plans, as well as for communicating with stakeholders. This great potential has already been recognized in the irrigation management sector, where a great number of test cases have already adopted the modelling paradigm of multi-agent simulation. Our current study of agent-based models for irrigation management draws some interesting conclusions, regarding the geographic and representation scale of the reviewed models, as well as the degree of stakeholder involvement in the various development phases. Overall, we argue that ABMS tools have a great potential in representing dynamic processes in integrated assessment tools for irrigation management. Such tools, when effectively capturing social interactions and coupling them with environmental and economical models, can promote active involvement of interested parties and produce sustainable and approvable solutions to irrigation management problems. @inproceedings{2007TzimaWASAMED, | |
Konstantinos N. Vavliakis, Andreas L. Symeonidis, Georgios T. Karagiannis and Pericles A. Mitkas
"Eikonomia-An Integrated Semantically Aware Tool for Description and Retrieval of Byzantine Art Information"
ICTAI, pp. 279--282, IEEE Computer Society, Washington, DC, USA, 2007 Oct
Semantic annotation and querying is currently applied on a number of versatile disciplines, providing the addedvalue of such an approach and, consequently the need for more elaborate \\\\96 either case-specific or generic \\\\96 tools. In this context, we have developed Eikonomia: an integrated semantically-aware tool for the description and retrieval of Byzantine Artwork Information. Following the needs of the ORMYLIA Art Diagnosis Center for adding semantics to their legacy data, an ontology describing Byzantine artwork based on CIDOCCRM, along with the interfaces for synchronization to and from the existing RDBMS have been implemented. This ontology has been linked to a reasoning tool, while a dynamic interface for the automated creation of semantic queries in SPARQL was developed. Finally, all the appropriate interfaces were instantiated, in order to allow easy ontology manipulation, query results projection and restrictions creation. @inproceedings{2007VavliakisICTAI, |
2006
Conference Papers
Andreas l. Symeonidis, Dionisis Kehagias, Adamantios Koumpis and Apostolos Vontas
"Open source supply chains"
10th ISPE International Conference on Concurrent Engineering (ISPE CE 2003), pp. 31--36, A. A. Balkema Publishers, Dubai/Sharjah, UAE, 2006 Jan
Enterprise Resource Planning (ERP) systems tend to deploy Supply Chains (SC), in order to successfully integrate customers, suppliers, manufacturers and warehouses, and therefore minimize systemwide costs while satisfying service level requirements. Although efficient, these systems are neither versatile nor adaptive, since newly discovered customer trends cannot be easily integrated. Furthermore, the development of such systems is conformed to strict licensing, since the exploitation of such kind of software is most of the times proprietary. This leads to a monolithic approach and to sub-utilization of efforts from all sides. Introducing a completely new paradigm of how primitive Supply Chain Management (SCM) rules apply on ERP systems, the developed framework is an Open Source MAS that introduces adaptive intelligence as a powerful add-on for ERP software customization. In this paper the SCM system developed is described, whereas the expected benefits of the open source initiative employed are illustrated. @inproceedings{2003SymeonidisISPE, | |
Z. Abas, Andreas L. Symeonidis, Alex andros Batzios, Zoi Basdagianni, Georgios Banos, Pericles A. Mitkas, E. Sinapis and A. Pampoukidou
"AMNOS-mobile: Exploiting handheld computers in efficient sheep recording"
35th ICAR, pp. 99--104, IEEE Computer Society, Kuopio, Finland, 2006 Jun
This paper focuses on AMNOS-mobile, a PDA application developed to support the tasks undertaken by sheep inspector when visiting the farms. It works in close cooperation with AMNOS, an integrated web-based platform developed to record, monitor, evaluate and manage the dairy sheep population of the Chios and Serres breed in Greece. Within the context of this paper, the design features of AMNOS-mobile are presented and the problems tackled by the use of handheld devices are discussed, illustrating how our application can enhance recording, improve the collection data process, and help farmers to more efficiently manage their flocks. @inproceedings{2006AbasICAR, | |
Chrysa Collyda, Sotiris Diplaris, Pericles A. Mitkas, N. Maglaveras and C. Pappas
"Fuzzy Hidden Markov Models: A New Approach In Multiple Sequence Alignment"
20th International Congress of the European Federation for Medical Informatics (MIE 2006) Stud Health Technol Inform, pp. 99--104, IEEE Computer Society, Maastricht, Netherlands, 2006 Aug
This paper proposes a novel method for aligning multiple genomic or proteomic sequences using a fuzzyfied Hidden Markov Model (HMM). HMMs are known to provide compelling performance among multiple sequence alignment (MSA) algorithms, yet their stochastic nature does not help them cope with the existing dependence among the sequence elements. Fuzzy HMMs are a novel type of HMMs based on fuzzy sets and fuzzy integrals which generalizes the classical stochastic HMM, by relaxing its independence assumptions. In this paper, the fuzzy HMM model for MSA is mathematically defined. New fuzzy algorithms are described for building and training fuzzy HMMs, as well as for their use in aligning multiple sequences. Fuzzy HMMs can also increase the model capability of aligning multiple sequences mainly in terms of computation time. Modeling the multiple sequence alignment procedure with fuzzy HMMs can yield a robust and time-effective solution that can be widely used in bioinformatics in various applications, such as protein classification, phylogenetic analysis and gene prediction, among others. @inproceedings{2006CollydaMIE, | |
Christos Dimou, Alexanros Batzios, Andreas L. Symeonidis and Pericles A. Mitkas
"A Multi-Agent Simulation Framework for Spiders Traversing the Semantic Web"
IEEE/WIC/ACM International Conference on Web Intelligence - WI 2006, pp. 736--739, Springer Berlin / Heidelberg, Hong Kong, China, 2006 Dec
Although search engines traditionally use spiders for traversing and indexing the web, there has not yet been any methodological attempt to model, deploy and test learning spiders. Moreover, the flourishing of the Semantic Web provides understandable information that may enhance search engines in providing more accurate results. Considering the above, we introduce BioSpider, an agent-based simulation framework for developing and testing autonomous, intelligent, semantically- focused web spiders. BioSpider assumes a direct analogy of the problem at hand with a multi-variate ecosystem, where each member is self-maintaining. The population of the ecosystem comprises cooperative spiders incorporating communication, mo- bility and learning skills, striving to improve efficiency. Genetic algorithms and classifier rules have been employed for spider adaptation and learning. A set of experiments has been set up in order to qualitatively test the efficacy and applicability of the proposed approach. @inproceedings{2006DimouWI, | |
Demetrios G. Eliades, Andreas L. Symeonidis and Pericles A. Mitkas
"GeneCity: A multi-agent simulation environment for hereditary diseases"
4th ACS/IEEE International Conference on Computer Systems and Applications - AICCSA 06, pp. 529--536, Springer-Verlag, Dubai/Sharjah, UAE, 2006 Mar
Simulating the psycho-societal aspects of a human com- munity is an issue always intriguing and challenging, as- piring us to help better understand, macroscopically, the way(s) humans behave. The mathematical models that have extensively been used for the analytical study of the vari- ous related phenomena prove inefficient, since they cannot conceive the notion of population heterogeneity, a parame- ter highly critical when it comes to community interactions. Following the more successful paradigm of artificial soci- eties, coupled with multi-agent systems and other Artificial Intelligence primitives, and extending previous epidemio- logical research work, we have developed GeneCity: an extended agent community, where agents live and interact under the veil of a hereditary epidemic. The members of the community, which can be either healthy, carriers of a trait, or patients, exhibit a number of human-like social (and medical) characteristics: wealth, acceptance and in- fluence, fear and knowledge, phenotype and reproduction ability. GeneCity provides a highly-configurable interface for simulating social environments and the way they are affected with the appearance of a hereditary disease, ei- ther Autosome or X-linked. This paper presents an ana- lytical overview of the work conducted and examines a test- hypothesis based on the spreading of Thalassaemia major. @inproceedings{2006EliadesAICCSA, | |
Dionisis Kehagias, Panos Toulis and Pericles A. Mitkas
"A Long-Term Profit Seeking Strategy for Continuous Double Auctions in a Trading Agent Competition"
Fourth Hellenic Conference on Artificial Intelligence, pp. 127--136, Springer-Verlag, Heraklion,Crete,Greece, 2006 May
The leap from decision support to autonomous systems has often raised a number of issues, namely system safety, soundness and security. Depending on the field of application, these issues can either be easily overcome or even hinder progress. In the case of Supply Chain Management (SCM), where system performance implies loss or profit, these issues are of high importance. SCM environments are often dynamic markets providing incomplete information, therefore demanding intelligent solutions which can adhere to environment rules, perceive variations, and act in order to achieve maximum revenue. Advancing on the way such autonomous solutions deal with the SCM process, we have built a robust, highly-adaptable and easily-configurable mechanism for efficiently dealing with all SCM facets, from material procurement and inventory management to goods production and shipment. Our agent has been crash-tested in one of the most challenging SCM environments, the trading agent competition SCM game and has proven capable of providing advanced SCM solutions on behalf of its owner. This paper introduces Mertacor and its main architectural primitives, provides an overview of the TAC SCM environment, and discusses Mertacor\\\\92s performance. @inproceedings{2006KehagiasHCAI, | |
Ioannis Kontogounnis, Kyriakos C. Chatzidimitriou, Andreas L. Symeonidis and Pericles A. Mitkas
"A Robust Agent Design for Dynamic SCM environments"
4th Hellenic Conference on Artificial Intelligence (SETN 06), pp. 127--136, Springer-Verlag, Heraklion, Crete, Greece, 2006 May
The leap from decision support to autonomous systems has often raised a number of issues, namely system safety, soundness and security. Depending on the field of application, these issues can either be easily overcome or even hinder progress. In the case of Supply Chain Management (SCM), where system performance implies loss or profit, these issues are of high importance. SCM environments are often dynamic markets providing incomplete information, therefore demanding intelligent solutions which can adhere to environment rules, perceive variations, and act in order to achieve maximum revenue. Advancing on the way such autonomous solutions deal with the SCM process, we have built a robust, highly-adaptable and easily-configurable mechanism for efficiently dealing with all SCM facets, from material procurement and inventory management to goods production and shipment. Our agent has been crash-tested in one of the most challenging SCM environments, the trading agent competition SCM game and has proven capable of providing advanced SCM solutions on behalf of its owner. This paper introduces Mertacor and its main architectural primitives, provides an overview of the TAC SCM environment, and discusses Mertacor\\\\92s performance. @inproceedings{2006KontogounnisSETN, | |
Pericles A. Mitkas, Anastasios N. Delopoulos, Andreas L. Symeonidis and Fotis E. Psomopoulos
"A Framework for Semantic Data Integration and Inferencing on Cervical Cancer"
Hellenic Bioinformatics and Medical Informatics Meeting, pp. 23-26, IEEE Computer Society, Biomedical Research Foundation, Academy of Athens, Greece, 2006 Oct
Advances in the area of biomedicine and bioengineering have allowed for more accurate and detailed data acquisition in the area of health care. Examinations that once were time- and cost-forbidding, are now available to public, providing physicians and clinicians with more patient data for diagnosis and successful treatment. These data are also used by medical researchers in order to perform association studies among environmental agents, virus characteristics and genetic attributes, extracting new and interesting risk markers which can be used to enhance early diagnosis and prognosis. Nevertheless, scientific progress is hindered by the fact that each medical center operates in relative isolation, regarding datasets and medical effort, since there is no universally accepted archetype/ontology for medical data acquisition, data storage and labeling. This, exactly, is the major goal of ASSIST: to virtually unify multiple patient record repositories, physically located at different laboratories, clinics and/or hospitals. ASSIST focuses on cervical cancer and implements a semantically-aware integration layer that unifies data in a seamless manner. Data privacy and security are ensured by techniques for data anonymization, secure data access and storage. Both the clinician as well as the medical researcher will have access to a knowledge base on cervical cancer and will be able to perform more complex and elaborate association studies on larger groups. @inproceedings{2006MitkasASSISTBioacademy, | |
Helen E. Polychroniadou, Fotis E. Psomopoulos and Pericles A. Mitkas
"G-Class: A Divide and Conquer Application for Grid Protein Classification"
Proceedings of the 2nd ADMKD 2006: Workshop on Data Mining and Knowledge Discovery (in conjunction with ADBIS 2006: The 10th East-European Conference on Advances in Databases and Information Systems), pp. 121-132, IEEE Computer Society, Thessaloniki, Greece, 2006 Sep
Protein classification has always been one of the major challenges in modern functional proteomics. The presence of motifs in protein chains can make the prediction of the functional behavior of proteins possible. The correlation between protein properties and their motifs is not always obvious, since more than one motif may exist within a protein chain. Due to the complexity of this correlation most data mining algorithms are either non efficient or time consuming. In this paper a data mining methodology that utilizes grid technologies is presented. First, data are split into multiple sets while preserving the original data distribution in each set. Then, multiple models are created by using the data sets as independent training sets. Finally, the models are combined to produce the final classification rules, containing all the previously extracted information. The methodology is tested using various protein and protein class subsets. Results indicate the improved time efficiency of our technique compared to other known data mining algorithms. @inproceedings{2006PolychroniadouGClass, | |
Fotis E. Psomopoulos and Pericles A. Mitkas
"PROTEAS: A Finite State Automata based data mining algorithm for rule extraction in protein classification"
Proceedings of the 5th Hellenic Data Management Symposium (HDMS), pp. 118-126, IEEE Computer Society, Thessaloniki, Greece, 2006 Sep
An important challenge in modern functional proteomics is the prediction of the functional behavior of proteins. Motifs in protein chains can make such a prediction possible. The correlation between protein properties and their motifs is not always obvious, since more than one motifs may exist within a protein chain. Thus, the behavior of a protein is a function of many motifs, where some overpower others. In this paper a data mining approach for a motif-based classification of proteins is presented. A new classification algorithm that induces rules and exploits finite state automata is introduced. First, data are modeled by terms of prefix tree acceptors, which are later merged into finite state automata. Finally, a new algorithm is proposed, for the induction of protein classification rules from finite state automata. The data mining model is trained and tested using various protein and protein class subsets, as well as the whole dataset of known proteins and protein classes. Results indicate the efficiency of our technique compared to other known data mining algorithms. @inproceedings{2006PsomopoulosHDMS, | |
Andreas L. Symeonidis, Vivia Nikolaidou and Pericles A. Mitkas
"Exploiting Data Mining Techniques for Improving the Efficiency of a Supply Chain Management Agent"
WI-IATW 06: Proceedings of the 2006 IEEE/WIC/ACM international conference on Web Intelligence and Intelligent Agent Technology, pp. 23-26, IEEE Computer Society, Hong Kong, China, 2006 Dec
Supply Chain Management (SCM) environments are often dynamic markets providing a plethora of information, either complete or incomplete. It is, therefore, evident that such environments demand intelligent solutions, which can perceive variations and act in order to achieve maximum revenue. To do so, they must also provide some sophisticated mechanism for exploiting the full potential of the environments they inhabit. Advancing on the way autonomous solutions usually deal with the SCM process, we have built a robust and highly-adaptable mechanism for efficiently dealing with all SCM facets, while at the same time incorporating a module that exploits data mining technology in order to forecast the price of the winning bid in a given order and, thus, adjust its bidding strategy. The paper presents our agent, Mertacor, and focuses on the forecasting mechanism it incorporates, aiming to optimal agent efficiency. @inproceedings{2006SymeonidisIADM, | |
Panos Toulis, Dionisis Kehagias and Pericles A. Mitkas
"Mertacor: A successful autonomous trading agent"
Autonomous Agents & Multi Agent Systems (AAMAS06), pp. 1191-1198, Springer-Verlag, Hakodate,Japan, 2006 May
The leap from decision support to autonomous systems has often raised a number of issues, namely system safety, soundness and security. Depending on the field of application, these issues can either be easily overcome or even hinder progress. In the case of Supply Chain Management (SCM), where system performance implies loss or profit, these issues are of high importance. SCM environments are often dynamic markets providing incomplete information, therefore demanding intelligent solutions which can adhere to environment rules, perceive variations, and act in order to achieve maximum revenue. Advancing on the way such autonomous solutions deal with the SCM process, we have built a robust, highly-adaptable and easily-configurable mechanism for efficiently dealing with all SCM facets, from material procurement and inventory management to goods production and shipment. Our agent has been crash-tested in one of the most challenging SCM environments, the trading agent competition SCM game and has proven capable of providing advanced SCM solutions on behalf of its owner. This paper introduces Mertacor and its main architectural primitives, provides an overview of the TAC SCM environment, and discusses Mertacor\\\\92s performance. @inproceedings{2006ToulisAAMAS, |
2005
Conference Papers
Ioannis N. Athanasiadis and Pericles A. Mitkas
"A distributed system for managing and diffusing environmental information"
5th International Exhibition and Conference on Environmental Technology (HELECO 05), Environment and Development (HELECO, pp. 422--428, ACTA Press, Athens, Greece, 2005 Feb
In an effort to support Environmental Monitoring and Surveillance Centers (EMSC) to fuse, manage and diffuse environmental data in a more efficient manner, we developed a distributed system for managing and diffusing environmental information. The developed system, called AISLE, is an adaptive, intelligent tool for supporting advanced information management services. Its main characteristic is the provision of decision support and information diffusion abilities through electronic services to several users with diverse needs. Specifically, software agents are in charge of integrating and managing environmental data recorded by field sensors or other monitoring devices, along with their diffusion to a wide range of end-user applications, such as environmental databases, terminal applications, or public information services over the internet. The system has been demonstrated in two pilot cases. In the first case, AISLE has been applied for assessing and reporting ambient air quality in Valencia, Spain. In the second case, AISLE was used for monitoring weather conditions in Cyprus. @inproceedings{2005AthanasiadisHELECO, | |
Ioannis N. Athanasiadis, A. K. Mentes, Pericles A. Mitkas and Yiannis A. Mylopoulos
"A system for evaluating water pricing alternatives in urban areas"
HELECO05, Water Management Section: New Legislative Framework for the Integrated Water Resources Management Track, Technical Chamber of Gree, 2005 Feb
@inproceedings{2005AthanasiadisHELECO05, | |
Ioannis N. Athanasiadis, Marios Milis, Pericles A. Mitkas and Silas C. Michaelides
"Abacus: A multi-agent system for meteorological radar data management and decision support"
Sixth Intl Symposium on Environmental Software Systems (ISESS05), pp. 183--187, Springer Berlin / Heidelberg, Sesimbra, Portugal, 2005 May
The continuous processing and evaluation of meteorological radar data require significant efforts by scientists, both for data processing, storage, and maintenance, and for data interpretation and visualization. To assist meteorologists and to automate a large part of these tasks, we have designed and developed Abacus, a multi-agent system for managing radar data and providing decision support. Abacus’ agents undertake data management and visualization tasks, while they are also responsible for extracting statistical indicators and assessing current weather conditions. Abacus agent system identifies potentially hazardous incidents, disseminates preprocessed information over the web, and enables warning services provided via email notifications. In this paper, Abacus’ agent architecture is detailed and agent communication for information diffusion is presented. Focus is also given on the customizable logical rule-bases for agent reasoning required in decision support. The platform has been tested with real-world data from the Meteorological Service of Cyprus. @inproceedings{2005AthanasiadisISESS, | |
Ioannis N. Athanasiadis, Andreas Solsbach, Pericles A. Mitkas and Jorge Marx Gómez
"An Agent-based Middleware for Environmental Information Management"
Second Symposium on Information Technologies in Environmental Engineering (ITEE 2005), pp. 1371-1374, ICSC-NAISO Academic Press, Magdeburg, Germany, 2005 Sep
Accurate protein classification is one of the major challenges in modern bioinformatics. Motifs that exist in the protein chain can make such a classification possible. A plethora of algorithms to address this problem have been proposed by both the artificial intelligence and the pattern recognition communities. In this paper, a data mining methodology for classification rules induction in proposed. Initially, expert – based protein families are processed to create a new hybrid set of families. Then, a prefix tree acceptor is created from the motifs in the protein chains, and subsequently transformed into a stochastic finite state automaton using the ALERGIA algorithm. Finally, an algorithm is presented for the extraction of classification rules from the automaton. @inproceedings{2005AthanasiadisITEE, | |
Chrysa Collyda, Sotiris Diplaris, Anastasios Delopoulos, Nikolaos Maglaveras, Pericles A. Mitkas and C. Pappas
"Towards building a model for the unification of distributed and heterogeneous biomedical repositories"
10th International Symposium for Health Information Management Research, Thessaloniki, Greece, 2005 Sep
@inproceedings{2005CollydaISHIMR, | |
Christos Dimou and Pericles A. Mitkas
"Biogrid: An Agent-based Metacomputing Ecosystem"
10th Panhellenic Conference on Informatics, pp. 88--98, Springer-Verlag, Volos, Greece, 2005 Nov
Nowadays, the number of protein sequences being stored in central protein databases from labs all over the world is constantly increasing. From these proteins only a fraction has been experimentally analyzed in order to detect their structure and hence their function in the corresponding organism. The reason is that experimental determination of structure is labor-intensive and quite time-consuming. Therefore there is the need for automated tools that can classify new proteins to structural families. This paper presents a comparative evaluation of several algorithms that learn such classification models from data concerning patterns of proteins with known structure. In addition, several approaches that combine multiple learning algorithms to increase the accuracy of predictions are evaluated. The results of the experiments provide insights that can help biologists and computer scientists design high-performance protein classification systems of high quality. @inproceedings{2005DimouPCI, | |
Sotiris Diplaris, Grigorios Tsoumakas, Pericles A. Mitkas and Ioannis Vlahavas
"Protein classification with multiple algorithms"
10th Panhellenic Conference in Informatics, pp. 448--456, Springer-Verlag, Volos, Greece, 2005 Nov
Nowadays, the number of protein sequences being stored in central protein databases from labs all over the world is constantly increasing. From these proteins only a fraction has been experimentally analyzed in order to detect their structure and hence their function in the corresponding organism. The reason is that experimental determination of structure is labor-intensive and quite time-consuming. Therefore there is the need for automated tools that can classify new proteins to structural families. This paper presents a comparative evaluation of several algorithms that learn such classification models from data concerning patterns of proteins with known structure. In addition, several approaches that combine multiple learning algorithms to increase the accuracy of predictions are evaluated. The results of the experiments provide insights that can help biologists and computer scientists design high-performance protein classification systems of high quality. @inproceedings{2005DiplarisPCI, | |
H. Eleftherohorinou, Sotiris ris, Pericles A. Mitkas and Georgios Banos
"AGELI: An Integrated Platform for the Assessment of National Genetic Evaluation Results by Learning and Informing"
Interbull Annual Meeting, pp. 183--187, Springer Berlin / Heidelberg, Uppsala, Sweden, 2005 Jun
One of the most interesting issues in agent technology has always been the modeling and enhancement of agent behavior. Numerous approaches exist, attempting to optimally reflect both the inner states, as well as the perceived environment of an agent, in order to provide it either with reactivity or proactivity. Within the context of this paper, an alternative methodology for enhancing agent behavior is presented. The core feature of this methodology is that it exploits knowledge extracted by the use of data mining techniques on historical data, data that describe the actions of agents within the MAS they reside. The main issues related to the design, development, and evaluation of such a methodology for predicting agent actions are discussed, while the basic concessions made to enable agent cooperation are outlined. We also present k-Profile, a new data mining mechanism for discovering action profiles and for providing recommendations on agent actions. Finally, indicative experimental results are apposed and discussed. @inproceedings{2005EleftherohorinouIAM, | |
Dionisis Kehagias and Pericles A. Mitkas
"Adaptive pricing functions for open outcry auctions"
Intelligent Agent Technology(IAT), pp. 653-656, IEEE Computer Society, Magdeburg, Germany, 2005 Sep
In agent-mediated marketplaces, autonomous agents deploy automated bidding mechanisms in order to increase revenue for humans. The ability of agents to estimate the next prices to be revealed in an auction, by applying forecasting, is a key element for efficient and successful bidding. In open outcry auctions, such as English and Dutch, information about bidders behavior is revealed at each round. This paper proposes a bid calculation function based on forecasting of the next price in English and Dutch auctions. The forecasting is based on two linear adaptive filters for stochastic estimation, whose parameters are calculated using a genetic algorithm. In order to test the efficiency of the two bidding methods and to benchmark the performance of the two filters, we conduct a set of experiments and present the results. @inproceedings{2005KehagiasIAT, | |
Pericles A. Mitkas
"Knowledge discovery for training intelligent agents"
International Work-shop on Autonomous Intelligent Systems: Agents and Data Mining (AIS-ADM 2005), pp. 161--174, Springer Berlin / Heidelberg, St.Petersburg,Russia, 2005 Jun
One of the most interesting issues in agent technology has always been the modeling and enhancement of agent behavior. Numerous approaches exist, attempting to optimally reflect both the inner states, as well as the perceived environment of an agent, in order to provide it either with reactivity or proactivity. Within the context of this paper, an alternative methodology for enhancing agent behavior is presented. The core feature of this methodology is that it exploits knowledge extracted by the use of data mining techniques on historical data, data that describe the actions of agents within the MAS they reside. The main issues related to the design, development, and evaluation of such a methodology for predicting agent actions are discussed, while the basic concessions made to enable agent cooperation are outlined. We also present k-Profile, a new data mining mechanism for discovering action profiles and for providing recommendations on agent actions. Finally, indicative experimental results are apposed and discussed. @inproceedings{2005MitakasAIS-ADM, | |
Pericles A. Mitkas, Andreas L. Symeonidis and Ioannis N. Athanasiadis
"A Retraining Methodology for Enhancing Agent Intelligence"
IEEE Intl Conference on Integration of Knowledge Intensive Multi-Agent Systems - KIMAS 05, pp. 422--428, Springer Berlin / Heidelberg, Waltham, MA, USA, 2005 Apr
Data mining has proven a successful gateway for discovering useful knowledge and for enhancing business intelligence in a range of application fields. Incorporating this knowledge into already deployed applications, though, is highly impractical, since it requires reconfigurable software architectures, as well as human expert consulting. In an attempt to overcome this deficiency, we have developed Agent Academy, an integrated development framework that supports both design and control of multi-agent systems (MAS), as well as ‘‘agent training’’. We define agent training as the automated incorporation of logic structures generated through data mining into the agents of the system. The increased flexibility and cooperation primitives of MAS, augmented with the training and retraining capabilities of Agent Academy, provide a powerful means for the dynamic exploitation of data mining extracted knowledge. In this paper, we present the methodology and tools for agent retraining. Through experimented results with the Agent Academy platform, we demonstrate how the extracted knowledge can be formulated and how retraining can lead to the improvement – in the long run – of agent intelligence. @inproceedings{2005MitkasKIMAS, | |
Fotis E. Psomopoulos and Pericles A. Mitkas
"A protein classification engine based on stochastic finite state automata"
Lecture Series on Computer and Computational Sciences VSP/Brill (Proceedings of the Symposium 35: Computational Methods in Molecular Biology in conjunction with ICCMSE), pp. 1371-1374, Springer-Verlag, Loutraki, Greece, 2005 Oct
Accurate protein classification is one of the major challenges in modern bioinformatics. Motifs that exist in the protein chain can make such a classification possible. A plethora of algorithms to address this problem have been proposed by both the artificial intelligence and the pattern recognition communities. In this paper, a data mining methodology for classification rules induction in proposed. Initially, expert – based protein families are processed to create a new hybrid set of families. Then, a prefix tree acceptor is created from the motifs in the protein chains, and subsequently transformed into a stochastic finite state automaton using the ALERGIA algorithm. Finally, an algorithm is presented for the extraction of classification rules from the automaton. @inproceedings{2005PsomopoulosICCMSE, | |
Andreas L. Symeonidis, Kyriakos C. Chatzidimitriou, Dionisis Kehagias and Pericles A. Mitkas
"An Intelligent Recommendation Framework for ERP Systems"
AIA 2005: Artificial Intelligence and Applications, pp. 422--428, ACTA Press, Innsbruck, Austria, 2005 Feb
Enterprise Resource Planning systems efficiently administer all tasks concerning real-time planning and manufacturing, material procurement and inventory monitoring, customer and supplier management. Nevertheless, the incorporation of domain knowledge and the application of adaptive decision making into such systems require extreme customization with a cost that becomes unaffordable, especially in the case of SMEs. We present an alternative approach for incorporating adaptive business intelligence into the company @inproceedings{2005SymeonidisAIA, | |
Andreas L. Symeonidis and Pericles A. Mitkas
"A Methodology for Predicting Agent Behavior by the Use of Data Mining Techniques"
Autonomous Intelligent Systems: Agents and Data Mining, pp. 161--174, Springer Berlin / Heidelberg, St. Petersburg, Russia, 2005 Jun
One of the most interesting issues in agent technology has always been the modeling and enhancement of agent behavior. Numerous approaches exist, attempting to optimally reflect both the inner states, as well as the perceived environment of an agent, in order to provide it either with reactivity or proactivity. Within the context of this paper, an alternative methodology for enhancing agent behavior is presented. The core feature of this methodology is that it exploits knowledge extracted by the use of data mining techniques on historical data, data that describe the actions of agents within the MAS they reside. The main issues related to the design, development, and evaluation of such a methodology for predicting agent actions are discussed, while the basic concessions made to enable agent cooperation are outlined. We also present k-Profile, a new data mining mechanism for discovering action profiles and for providing recommendations on agent actions. Finally, indicative experimental results are apposed and discussed. @inproceedings{2005SymeonidisAISADM, |
2004
Conference Papers
Ioannis N. Athanasiadis and Pericles A. Mitkas
"Software Agents for Assessing Environmental Quality: Advantages and Limitations"
18th International Conference Informatics for Environmental Protection: Sharing (EnviroInfo 2004), Geneva, Switzerland, 2004 Oct
@inproceedings{2004AthanasiadisICIEP, | |
Ioannis N. Athanasiadis and Pericles A. Mitkas
"Applying agent technology in environmental management systems under real-time constraints"
Second Biennial Meeting of the International Environmental Modelling and Software Society at Environmental Informatics towards Citizen-centered Electronic Information Services Workshop, pp. 54--60, Osnabruck, Germany, 2004 Jun
Changes in the natural environment affect our quality of life. Thus, government, industry, and the public call for integrated environmental management systems capable of supplying all parties with validated, accurate and timely information. The ‘near real-time’ constraint reveals two critical problems in delivering such tasks: the low quality or absence of data, and the changing conditions over a long period. These problems are common in environmental monitoring networks and although harmless for off-line studies, they may be serious for near real-time systems. In this work, we discuss the problem space of near real-time reporting Environmental Management Systems and present a methodology for applying agent technology this area. The proposed methodology applies powerful tools from the IT sector, such as software agents and machine learning, and identifies the potential use for solving real-world problems. An experimental agent-based prototype developed for monitoring and assessing air-quality in near real time is presented. A community of software agents is assigned to monitor and validate measurements coming from several sensors, to assess air-quality, and, finally, to deliver air quality indicators and alarms to appropriate recipients, when needed, over the web. The architecture of the developed system is presented and the deployment of a real-world test case is demonstrated. @inproceedings{2004AthanasiadisIEMSS, | |
Ioannis N. Athanasiadis and Pericles A. Mitkas
"Supporting the Decision-Making Process in Environmental Monitoring Systems with Knowledge Discovery Techniques"
KDnet Symposium Knowledge Discovery for Environmental Management, pp. 1--12, Bonn, Germany, 2004 Jun
In this paper an empirical approach for supporting the decision making process involved in an Environmental Management System (EMS) that monitors air quality and triggers air quality alerts is presented. Data uncertainty problems associated with an air quality monitoring network, such as measurement validation and estimation of missing or erroneous values, are addressed through the exploitation of data mining techniques. Exhaustive experiments with real world data have produced trustworthy predictive models, capable of supporting the decision-making process. The outstanding performance of the induced predictive models indicate the added value of this approach for supporting the decision making process in an EMS. @inproceedings{2004AthanasiadisSKDEM, | |
Sotiris Diplaris, Andreas Symeonidis, Pericles A. Mitkas, Georgios Banos and Z. Abas
"An Alarm Firing System for National Genetic Evaluation Quality Control"
Interbull Annual Meeting, pp. 146--150, Tunis, Tunisia, 2004 May
In this paper an empirical approach for supporting the decision making process involved in an Environmental Management System (EMS) that monitors air quality and triggers air quality alerts is presented. Data uncertainty problems associated with an air quality monitoring network, such as measurement validation and estimation of missing or erroneous values, are addressed through the exploitation of data mining techniques. Exhaustive experiments with real world data have produced trustworthy predictive models, capable of supporting the decision-making process. The outstanding performance of the induced predictive models indicate the added value of this approach for supporting the decision making process in an EMS. @inproceedings{2004DiplarisIAM, | |
D. Kehagias, Kyriakos C. Chatzidimitriou, Andreas Symeonidis and Pericles A. Mitkas
"Information Agents Cooperating with Heterogeneous Data Sources for Customer-Order Management"
Paper presented at the 19th Annual ACM Symposium on Applied Computing (SAC 2004), pp. 52--57, Nicosia, Cyprus, 2004 Mar
As multi-agent systems and information agents obtain an in- creasing acceptance by application developers, existing legacy Enterprise Resource Planning (ERP) systems still provide the main source of data used in customer, supplier and inventory resource management. In this paper we present a multi-agent system, comprised of information agents, which cooperates with a legacy ERP in order to carry out orders posted by customers in an enterprise environment. Our system is enriched by the capability of producing recommendations to the interested customer through agent cooperation. At first, we address the problem of information workload in an enterprise environment and explore the opportunity of a plausible solution. Secondly we present the architecture of our system and the types of agents involved in it. Finally, we show how it manipulates retrieved information for efficient and facile customer-order management and illustrate results derived from real-data. @inproceedings{2004KehagiasSAC, | |
Fotis E. Psomopoulos, Sotiris Diplaris and Pericles A. Mitkas
"A finite state automata based technique for protein classification rules induction"
Proceedings of the Second European Workshop on Data Mining and Text Mining in Bioinformatics (in conjunction with ECML/PKDD), pp. 54--60, Pisa, Italy, 2004 Sep
An important challenge in modern functional proteomics is the prediction of the functional behavior of proteins. Motifs in protein chains can make such a prediction possible. The correlation between protein properties and their motifs is not always obvious, since more than one motifs can exist within a protein chain. Thus, the behavior of a protein is a function of many motifs, where some overpower others. In this paper a data-mining approach for motif-based classification of proteins is presented. A new classification rules inducing algorithm that exploits finite state automata is introduced. First, data are modeled by terms of prefix tree acceptors, which are later merged into finite state automata. Finally, we propose a new algorithm for the induction of protein classification rules from finite state automata. The data-mining model is trained and tested using various protein and protein class subsets, as well as the whole dataset of known proteins and protein classes. Results indicate the efficiency of our technique compared to other known data-mining algorithms. @inproceedings{2004PsomopoulosPSEWDMTMB, |
2003
Conference Papers
Ioannis N. Athanasiadis, Pericles A. Mitkas, G. B. Laleci and Y. Kabak
"Embedding data-driven decision strategies on software agents: The case of a Multi-Agent System for Monitoring Air-Quality Indexes"
10th ISPE International Conference on Concurrent Engineering: Research and Applications, pp. 23--30, Madeira, Portugal, 2003 Jul
This paper describes the design and deployment of an agent community, which is responsible for monitoring and assessing air quality, based on measurements generated by a meteorological station. Software agents acting as mediators or decision makers deliver validated information to the appropriate destinations. We outline the procedure for creating agent ontologies, agent types, and, finally, for training agents based on his- torical data volumes. The C4.5 algorithm for decision tree extraction is applied on meteorological and air-pollutant measurements. The decision models extracted are related to the validation of incoming measurements and to the estimation of missing or erroneous measurements. Emphasis is given on the agent training process, which must embed these data-driven decision models on software agents in a simple and effortless way. We developed a prototype system, which demonstrates the advantages of agent-based solutions for intelligent environmental applications. @inproceedings{2003AthanasiadisISPE, | |
Ioannis N. Athanasiadis, V. G. Kaburlasos, Pericles A. Mitkas and V. Petridis
"Applying Machine Learning Techniques on Air Quality Data for Real-Time Decision Support"
First International NAISO Symposium on Information Technologies in Environmental Engineering (ITEE 2003), pp. 11--18, Gdansk, Poland, 2003 Jun
Fairly rapid environmental changes call for continuous surveillance and decision making, areas where IT technologies can be valuable. In the aforementioned context this work describes the application of a novel classifier, namely ?-FLNMAP, for estimating the ozone concentration level in the atmosphere. In a series of experiments on meteorological and air pollutants data, the ?–FLNMAP classifier compares favorably with both back-propagation neural networks and the C4.5 algorithm; moreover ?–FLNMAP induces only a few rules from the data. The ?–FLNMAP classifier can be implemented as either a neural network or a decision tree. We also discuss the far reaching potential of ?–FLNMAP in IT applications due to its applicability on partially (lattice) ordered data. @inproceedings{2003AthanasiadisITEE, | |
Sotiris Diplaris, Andreas L. Symeonidis, Pericles A. Mitkas, Georgios Banos and Z. Abas
"Quality Control of National Genetic Eva luation Results Using Data-Mining Techniques; A Progress Report"
Interbull Annual Meeting, pp. 8--15, Rome, Italy, 2003 Aug
The continuous expansion of Internet has enabled the development of a wide range of advanced digital services. Real-time data diffusion has elimi- nated processing bottlenecks and has led to fast, easy and no-cost communications. This primitive has been widely exploited in the process of job searching. Numerous systems have been developed offering job candidates with the opportunity to browse for vacancies, submit resumes, and even contact the most appealing of the employers. Although effective, most of these systems are characterized by their simplicity, acting more like an enhanced bulletin board, rather than an integrated, fully functional system. Even for the more advanced of these systems user interaction is obligatory, in order to couple job seekers with job providers, thus continuous supervising of the process is unavoidable. Advancing on the way primitive job recruitment techniques apply on Internet-based systems, and dealing with their lack of efficiency and interactivity, we have de- veloped a robust software system that employs intelligent techniques for coupling candidates and jobs, according to the formers’ skills and the latter’s requirements. A thorough analysis of the system specifications has been conducted, and all issues concerning information retrieval and data filtering, coupling intelligence, storage, security, user interaction and ease-of-use have been integrated into one web-based job portal @inproceedings{2003BanosIAM, | |
Gerasimos Hatzidamianos, Sotiris Diplaris, Ioannis N. Athanasiadis and Pericles A. Mitkas
"GenMiner: A data mining tool for protein analysis"
9th Panhellenic Conference in Informatics, pp. 346--360, Thessaloniki, Greece, 2003 Nov
We present an integrated tool for preprocessing and analysis of genetic data through data mining. Our goal is the prediction of the functional behavior of proteins, a critical problem in functional genomics. During the last years, many programming approaches have been developed for the identification of short amino-acid chains, which are included in families of related proteins. These chains are called motifs and they are widely used for the prediction of the protein’s behavior, since the latter is dependent on them. The idea to use data mining techniques stems from the sheer size of the problem. Since every protein consists of a specific number of motifs, some stronger than others, the identification of the properties of a protein requires the examination of immeasurable combinations. The presence or absence of stronger motifs affects the way in which a protein reacts. GenMiner is a preprocessing software tool that can receive data from three major protein databases and transform them in a form suitable for input to the WEKA data mining suite. A decision tree model was created using the derived training set and an efficiency test was conducted. Finally, the model was applied to unknown proteins. Our experiments have shown that the use of the decision tree model for mining protein data is an efficient and easy-to-implement solution, since it possesses a high degree of parameterization and therefore, it can be used in a plethora of cases. @inproceedings{2003HatzidamianosPCI, | |
G. Milis, Andreas L. Symeonidis and Pericles A. Mitkas
"Ergasiognomon: A Model System of Advanced Digital Services Designed and Developed to Support the Job Marketplace"
9th Panhellenic Conference in Informatics, pp. 346--360, Thessaloniki, Greece, 2003 Nov
The continuous expansion of Internet has enabled the development of a wide range of advanced digital services. Real-time data diffusion has elimi- nated processing bottlenecks and has led to fast, easy and no-cost communications. This primitive has been widely exploited in the process of job searching. Numerous systems have been developed offering job candidates with the opportunity to browse for vacancies, submit resumes, and even contact the most appealing of the employers. Although effective, most of these systems are characterized by their simplicity, acting more like an enhanced bulletin board, rather than an integrated, fully functional system. Even for the more advanced of these systems user interaction is obligatory, in order to couple job seekers with job providers, thus continuous supervising of the process is unavoidable. Advancing on the way primitive job recruitment techniques apply on Internet-based systems, and dealing with their lack of efficiency and interactivity, we have de- veloped a robust software system that employs intelligent techniques for coupling candidates and jobs, according to the formers’ skills and the latter’s requirements. A thorough analysis of the system specifications has been conducted, and all issues concerning information retrieval and data filtering, coupling intelligence, storage, security, user interaction and ease-of-use have been integrated into one web-based job portal @inproceedings{2003MilisPCI, | |
Pericles A. Mitkas, Dionisis Kehagias, Andreeas L. Symeonidis and I. N. Athanasiadis
"A Framework for Constructing Multi-Agent Applications and Training Intelligent Agents"
4th International Workshop on Agent-Oriented Software Engineering (AOSE-2003), Autonomous Agents \& Multi-Agent Systems (AAMAS 2003), pp. 96--109, Melbourne, Australia, 2003 Jun
As agent-oriented paradigm is reaching a significant level of acceptance by software developers, there is a lack of integrated high-level abstraction tools for the design and development of agent-based applications. In an effort to mitigate this deficiency, we introduce Agent Academy, an integrated development framework, implemented itself as a multi-agent system, that supports, in a single tool, the design of agent behaviours and reusable agent types, the definition of ontologies, and the instantiation of single agents or multi-agent communities. In addition to these characteristics, our framework goes deeper into agents, by implementing a mechanism for embedding rule-based reasoning into them. We call this procedure «agent training» and it is realized by the application of AI techniques for knowledge discovery on application-specific data, which may be available to the agent developer. In this respect, Agent Academy provides an easy-to-use facility that encourages the substitution of existing, traditionally developed applications by new ones, which follow the agent-orientation paradigm. @inproceedings{2003MitkasAOSE, | |
Pericles A. Mitkas, Dionisis Kehagias, Andreas L. Symeonidis and Ioannis N. Athanasiadis
"Agent Academy: An integrated tool for developing multi-agent systems and embedding decision structures into agents"
First European Workshop on Multi-Agent Systems (EUMAS 2003), Oxford, UK, 2003 Dec
In this paper we present Agent Academy, a framework that enables software developers to quickly develop multi-agent applications, when prior historical data relevant to a desired rule-based behaviour are available. Agent Academy is implemented itself as a multi-agent system, that supports, in a single tool, the design of agent behaviours and reusable agent types, the definition of ontologies, and the instantiation of single agents or multi-agent communities. Once an agent has been designed within the framework, the agent developer can create a specific ontology that describes the historical data. In this way, agents become capable of having embedded rule-based reasoning. We call this procedure «agent training» and it is realized by the application of data mining and knowledge discovery techniques on the application-specific historical data. From this point of view, Agent Academy provides a tool for both creating multi-agent systems and embedding rule-based decision structures into one or more of the participating agents. @inproceedings{2003MitkasEUMAS, | |
Pericles A. Mitkas, Andreas Symeonidis, Dionisis Kehagias and Ioannis N. Athanasiadis
"Application of Data Mining and Intelligent Agent Technologies to Concurrent Engineering"
10th ISPE International Conference on Concurrent Engineering: Research and Applications, pp. 11--18, Madeira, Portugal, 2003 Jul
Software agent technology has matured enough to produce intelligent agents, which can be used to control a large number of Concurrent Engineering tasks. Multi-Agent Systems (MAS) are communities of agents that exchange information and data in the form of messages. The agents intelligence can range from rudimentary sensor monitoring and data reporting, to more advanced forms of decision-making and autonomous behaviour. The behaviour and intelligence of each agent in the community can be obtained by performing Data Mining on available application data and the respected knowledge domain. We have developed Agent Academy (AA), a software platform for the design, creation, and deployment of MAS, which combines the power of knowledge discovery algorithms with the versatility of agents. Using this platform, we illustrate how agents, equipped with a data-driven inference engine, can be dynamically and continuously trained. We also discuss three prototype MAS developed with AA. @inproceedings{2003MitkasISPE, |
2002
Conference Papers
Z. Abas, G. Banos, Pericles A. Mitkas, P. Saragiotis and I. Maltaris
"AMNOS: An Integrated Web-Based Platform for Dairy Sheep Breeding Management"
7th World Congress on Genetics Applied to Livestock Production, pp. 757--764, Montpellier, France, 2002 Aug
The objective of this paper is to describe AMNOS, an integrated web-based platform, developed to record, monitor, evaluate and manage the dairy sheep population of the Chios breed in Greece. The key component of the platform is a database with several relations operating at the flock and individual animal level. The system is based on the Microsoft SQL server. Dynamic web pages are generated using the Microsoft ActiveX Data Object technology. Business logic was implemented on ASP pages, which are also responsible for creating the HTML pages sent to the user\\'s browser. A series of conventions and rules have been added in order to ensure incoming data integrity. The key advantages of AMNOS are accessibility and ease of management. User (sheep producer) participation is being currently solicited amongst members of the Breeders Cooperative that administers the system. @inproceedings{2002AbasWCGALP, | |
Dionisis Kehagias, Andreas L. Symeonidis, Pericles A. Mitkas and M. Alborg
"Towards improving Multi-Agent Simulation in safety management and hazard control environments"
Simulation and Planning in High Autonomy Systems AIS 2002, pp. 757--764, Lisbon, Portugal, 2002 Apr
This paper introduces the capabilities of Agent Academy in the area of Safety Management and Hazard Control Systems. Agent Academy is a framework under development, which uses data mining techniques for training intelligent agents. This framework generates software agents with an initial degree of intelligence and trains them to manipulate complex tasks. The agents, are further integrated into a simulation multi-agent environment capable of managing issues in a hazardous environment, as well as regulating the parameters of the safety management strategy to be deployed in order to control the hazards. The initially created agents take part in long agentto -agent transactions and their activities are formed into behavioural data, which are stored in a database. As soon as the amount of collected data increases sufficiently, a data mining process is initiated, in order to extract specific trends adapted by agents and improve their intelligence. The result of the overall procedure aims to improve the simulation environment of safety management. The communication of agents as well as the architectural characteristics of the simulation environment adheres to the set of specifications imposed by the Foundation for Intelligent Physical Agents (FIPA). @inproceedings{2002KehagiasAIS, | |
Pericles A. Mitkas, Andreas L. Symeonidis, Dionisis Kechagias, Ioannis N. Athanasiadis, G. Laleci, G. Kurt, Y. Kabak, A. Acar and A. Dogac
"An Agent Framework for Dynamic Agent Retraining: Agent Academy"
eBusiness and eWork 2002 (e2002) 12th annual conference and exhibition, pp. 757--764, Prague, Czech Republic, 2002 Oct
Agent Academy (AA) aims to develop a multi-agent society that can train new agents for specific or general tasks, while constantly retraining existing agents in a recursive mode. The system is based on collecting information both from the environment and the behaviors of the acting agents and their related successes/failures to generate a body of data, stored in the Agent Use Repository, which is mined by the Data Miner module, in order to generate useful knowledge about the application domain. Knowledge extracted by the Data Miner is used by the Agent Training Module as to train new agents or to enhance the behavior of agents already running. In this paper the Agent Academy framework is introduced, and its overall architecture and functionality are presented. Training issues as well as agent ontologies are discussed. Finally, a scenario, which aims to provide environmental alerts to both individuals and public authorities, is described an AA-based use case. @inproceedings{2002MitkaseBusiness, | |
Andreas Symeonidis, Pericles A. Mitkas and Dionisis Kehagias
"Mining Patterns and Rules for Improving Agent Intelligence Through an Integrated Multi-Agent Platform"
6th IASTED International Conference on Artificial Intelligence and Soft Computing (ASC 2002), pp. 757--764, Banff, Alberta, Canada, 2002 Jan
This paper introduces the capabilities of Agent Academy in the area of Safety Management and Hazard Control Systems. Agent Academy is a framework under development, which uses data mining techniques for training intelligent agents. This framework generates software agents with an initial degree of intelligence and trains them to manipulate complex tasks. The agents, are further integrated into a simulation multi-agent environment capable of managing issues in a hazardous environment, as well as regulating the parameters of the safety management strategy to be deployed in order to control the hazards. The initially created agents take part in long agentto -agent transactions and their activities are formed into behavioural data, which are stored in a database. As soon as the amount of collected data increases sufficiently, a data mining process is initiated, in order to extract specific trends adapted by agents and improve their intelligence. The result of the overall procedure aims to improve the simulation environment of safety management. The communication of agents as well as the architectural characteristics of the simulation environment adheres to the set of specifications imposed by the Foundation for Intelligent Physical Agents (FIPA). @inproceedings{2002SymeonidisASC, |