IFIP Advances in Information and Communication Technology
364
Editor-in-Chief A. Joe Turner, Seneca, SC, USA
Editorial Board Foundations of Computer Science Mike Hinchey, Lero, Limerick, Ireland Software: Theory and Practice Bertrand Meyer, ETH Zurich, Switzerland Education Arthur Tatnall, Victoria University, Melbourne, Australia Information Technology Applications Ronald Waxman, EDA Standards Consulting, Beachwood, OH, USA Communication Systems Guy Leduc, Université de Liège, Belgium System Modeling and Optimization Jacques Henry, Université de Bordeaux, France Information Systems Jan Pries-Heje, Roskilde University, Denmark Relationship between Computers and Society Jackie Phahlamohlaka, CSIR, Pretoria, South Africa Computer Systems Technology Paolo Prinetto, Politecnico di Torino, Italy Security and Privacy Protection in Information Processing Systems Kai Rannenberg, Goethe University Frankfurt, Germany Artificial Intelligence Tharam Dillon, Curtin University, Bentley, Australia Human-Computer Interaction Annelise Mark Pejtersen, Center of Cognitive Systems Engineering, Denmark Entertainment Computing Ryohei Nakatsu, National University of Singapore
IFIP – The International Federation for Information Processing IFIP was founded in 1960 under the auspices of UNESCO, following the First World Computer Congress held in Paris the previous year. An umbrella organization for societies working in information processing, IFIP’s aim is two-fold: to support information processing within ist member countries and to encourage technology transfer to developing nations. As ist mission statement clearly states, IFIP’s mission is to be the leading, truly international, apolitical organization which encourages and assists in the development, exploitation and application of information technology for the benefit of all people. IFIP is a non-profitmaking organization, run almost solely by 2500 volunteers. It operates through a number of technical committees, which organize events and publications. IFIP’s events range from an international congress to local seminars, but the most important are: • The IFIP World Computer Congress, held every second year; • Open conferences; • Working conferences. The flagship event is the IFIP World Computer Congress, at which both invited and contributed papers are presented. Contributed papers are rigorously refereed and the rejection rate is high. As with the Congress, participation in the open conferences is open to all and papers may be invited or submitted. Again, submitted papers are stringently refereed. The working conferences are structured differently. They are usually run by a working group and attendance is small and by invitation only. Their purpose is to create an atmosphere conducive to innovation and development. Refereeing is less rigorous and papers are subjected to extensive group discussion. Publications arising from IFIP events vary. The papers presented at the IFIP World Computer Congress and at open conferences are published as conference proceedings, while the results of the working conferences are often published as collections of selected and edited papers. Any national society whose primary activity is in information may apply to become a full member of IFIP, although full membership is restricted to one society per country. Full members are entitled to vote at the annual General Assembly, National societies preferring a less committed involvement may apply for associate or corresponding membership. Associate members enjoy the same benefits as full members, but without voting rights. Corresponding members are not represented in IFIP bodies. Affiliated membership is open to non-national societies, and individual and honorary membership schemes are also offered.
Lazaros Iliadis Ilias Maglogiannis Harris Papadopoulos (Eds.)
Artificial Intelligence Applications and Innovations 12th INNS EANN-SIG International Conference, EANN 2011 and 7th IFIP WG 12.5 International Conference, AIAI 2011 Corfu, Greece, September 15-18, 2011 Proceedings , Part II
13
Volume Editors Lazaros Iliadis Democritus University of Thrace 68200 N. Orestiada, Greece E-mail:
[email protected] Ilias Maglogiannis University of Central Greece 35100 Lamia, Greece E-mail:
[email protected] Harris Papadopoulos Frederick University 1036 Nicosia, Cyprus E-mail:
[email protected]
ISSN 1868-4238 e-ISSN 1868-422X e-ISBN 978-3-642-23960-1 ISBN 978-3-642-23959-5 DOI 10.1007/978-3-642-23960-1 Springer Heidelberg Dordrecht London New York Library of Congress Control Number: 2011935861 CR Subject Classification (1998): I.2, H.3, H.4, F.1, I.4, I.5
© IFIP International Federation for Information Processing 2011 This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation, broadcasting, reproduction on microfilms or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in ist current version, and permission for use must always be obtained from Springer. Violations are liable to prosecution under the German Copyright Law. The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Typesetting: Camera-ready by author, data conversion by Scientific Publishing Services, Chennai, India Printed on acid-free paper Springer is part of Springer Science+Business Media (www.springer.com)
Preface
Artificial intelligence (AI) is a rapidly evolving area that offers sophisticated and advanced approaches capable of tackling complicated and challenging problems. Transferring human knowledge into analytical models and learning from data is a task that can be accomplished by soft computing methodologies. Artificial neural networks (ANN) and support vector machines are two cases of such modeling techniques that stand behind the idea of learning. The 2011 co-organization of the 12th Engineering Applications of Neural Networks (EANN) and of the 7th Artificial Intelligence Applications and Innovations (AIAI) conferences was a major technical event in the fields of soft computing and AI, respectively. The first EANN was organized in Otaniemi, Finland, in 1995. It has had a continuous presence as a major European scientific event. Since 2009 it has been guided by a Steering Committee that belongs to the “EANN Special Interest Group” of the International Neural Network Society (INNS). The 12th EANN 2011 was supported by the INNS and by the IEEE branch of Greece. Moreover, the 7th AIAI 2011 was supported and sponsored by the International Federation for Information Processing (IFIP). The first AIAI was held in Toulouse, France, in 2004 and since then it has been held annually offering scientists the chance to present the achievements of AI applications in various fields. It is the official conference of the Working Group 12.5 “Artificial Intelligence Applications” of the IFIP Technical Committee 12, which is active in the field of AI. IFIP was founded in 1960 under the auspices of UNESCO, following the first World Computer Congress held in Paris the previous year. It was the first time ever that these two well-established events were hosted under the same umbrella, on the beautiful Greek island of Corfu in the Ionian Sea and more specifically in the Department of Informatics of the Ionian University. This volume contains the papers that were accepted to be presented orally at the 7th AIAI conference and the papers accepted for the First International Workshop on Computational Intelligence in Software Engineering (CISE) and the Artificial Intelligence Applications in Biomedicine (AIAB) workshops. The conference was held during September 15–18, 2011. The diverse nature of papers presented demonstrates the vitality of neural computing and related soft computing approaches and it also proves the very wide range of AI applications. On the other hand, this volume contains basic research papers, presenting variations and extensions of several approaches. The response to the call for papers was more than satisfactory with 150 papers initially submitted. All papers passed through a peer-review process by at least two independent academic referees. Where needed a third referee was consulted to resolve any conflicts. In the EANN/AIAI 2011 event, 34% of the submitted manuscripts (totally 52) were accepted as full papers, whereas 21% were ac-
VI
Preface
cepted as short ones and 45% (totally 67) of the submissions were rejected. The authors of accepted papers came from 27 countries all over Europe (e.g., Austria, Bulgaria, Cyprus, Czech Republic, Finland, France, Germany, Greece, Italy, Poland, Portugal, Slovakia, Slovenia, Spain, UK), America (e.g., Brazil, Canada, Chile, USA), Asia (e.g., China, India, Iran, Japan, Taiwan), Africa (e.g., Egypt, Tunisia) and Oceania (New Zealand). Three keynote speakers were invited and they gave lectures on timely aspects of AI and ANN. 1. Nikola Kasabov. Founding Director and Chief Scientist of the Knowledge Engineering and Discovery Research Institute (KEDRI), Auckland (www.kedri. info/). He holds a Chair of Knowledge Engineering at the School of Computing and Mathematical Sciences at Auckland University of Technology. He is a Fellow of the Royal Society of New Zealand, Fellow of the New Zealand Computer Society and a Senior Member of IEEE. He was Past President of the International Neural Network Society (INNS) and a Past President of the Asia Pacific Neural Network Assembly (APNNA). Title of the keynote presentation: “Evolving, Probabilistic Spiking Neural Network Reservoirs for Spatio- and Spectro-Temporal Data.” 2. Tom Heskes. Professor of Artificial Intelligence and head of the Machine Learning Group at the Institute for Computing and Information Sciences, Radboud University Nijmegen, The Netherlands. He is Principal Investigator at the Donders Centre for Neuroscience and Director of the Institute for Computing and Information Sciences. Title of the keynote presentation: “Reading the Brain with Bayesian Machine Learning.” 3. A.G. Cohn. Professor of Automated Reasoning, Director of Institute for Artificial Intelligence and Biological Systems, School of Computing, University of Leeds, UK. Tony Cohn holds a Personal Chair at the University of Leeds, where he is Professor of Automated Reasoning. He is presently Director of the Institute for Artificial Intelligence and Biological Systems. He leads a research group working on knowledge representation and reasoning with a particular focus on qualitative spatial/spatio-temporal reasoning, the best known being the well-cited region connection calculus (RCC). Title of the keynote presentation: “Learning about Activities and Objects from Video.” The EANN/AIAI conference consisted of the following main thematic sessions: – – – – – – – – – – –
AI in Finance, Management and Quality Assurance Computer Vision and Robotics Classification-Pattern Recognition Environmental and Earth Applications of AI Ethics of AI Evolutionary Algorithms—Optimization Feature Extraction-Minimization Fuzzy Systems Learning—Recurrent and RBF ANN Machine Learning and Fuzzy Control Medical Applications
Preface
– – – – –
VII
Multi-Layer ANN Novel Algorithms and Optimization Pattern Recognition-Constraints Support Vector Machines Web-Text Mining and Semantics
We would very much like to thank Hassan Kazemian (London Metropolitan University) and Pekka Kumpulainen (Tampere University of Technology, Finland) for their kind effort to organize successfully the Applications of Soft Computing to Telecommunications Workshop (ASCOTE). Moreover, we would like to thank Efstratios F. Georgopoulos (TEI of Kalamata, Greece), Spiridon Likothanassis, Athanasios Tsakalidis and Seferina Mavroudi (University of Patras, Greece) as well as Grigorios Beligiannis (University of Western Greece) and Adam Adamopoulos (Democritus University of Thrace, Greece) for their contribution to the organization of the Computational Intelligence Applications in Bioinformatics (CIAB) Workshop. We are grateful to Andreas Andreou (Cyprus University of Technology) and Harris Papadopoulos (Frederick University of Cyprus) for the organization of the Computational Intelligence in Software Engineering Workshop (CISE). The Artificial Intelligence Applications in Biomedicine (AIAB) Workshop was organized successfully in the framework of the 12th EANN 2011 conference and we wish to thank Harris Papadopoulos, Efthyvoulos Kyriacou (Frederick University of Cyprus) Ilias Maglogiannis (University of Central Greece) and George Anastassopoulos (Democritus University of Thrace, Greece). Finally, the Second Workshop on Informatics and Intelligent Systems Applications for Quality of Life information Services (2nd ISQLIS) was held successfully and we would like to thank Kostas Karatzas (Aristotle University of Thessaloniki, Greece) Lazaros Iliadis (Democritus University of Thrace, Greece) and Mihaela Oprea (University Petroleum-Gas of Ploiesti, Romania). The accepted papers of all five workshops (after passing through a peerreview process by independent academic referees) were published in the Springer proceedings. They include timely applications and theoretical research on specific subjects. We hope that all of them will be well established in the future and that they will be repeated every year in the framework of these conferences. We hope that these proceedings will be of major interest for scientists and researchers world wide and that they will stimulate further research in the domain of artificial neural networks and AI in general. September 2011
Dominic Palmer Brown
Organization
Executive Committee Conference Chair Dominic Palmer Brown
London Metropolitan University, UK
Program Chair Lazaros Iliadis Elias Maglogiannis Harris Papadopoulos
Democritus University of Thrace, Greece University of Central Greece, Greece Frederick University, Cyprus
Organizing Chair Vassilis Chrissikopoulos Yannis Manolopoulos Tutorials Michel Verleysen Dominic Palmer-Brown Chrisina Jayne Vera Kurkova
Ionian University, Greece Aristotle University, Greece
Universite catholique de Louvain, Belgium London Metropolitan University, UK London Metropolitan University, UK Academy of Sciences of the Czech Republic, Czech Republic
Workshops Hassan Kazemian Pekka Kumpulainen Kostas Karatzas Lazaros Iliadis Mihaela Oprea Andreas Andreou Harris Papadopoulos Spiridon Likothanassis Efstratios Georgopoulos Seferina Mavroudi Grigorios Beligiannis
London Metropolitan University, UK Tampere University of Technology, Finland Aristotle University of Thessaloniki, Greece Democritus University of Thrace, Greece University Petroleum-Gas of Ploiesti, Romania Cyprus University of Technology, Cyprus Frederick University, Cyprus University of Patras, Greece Technological Educational Institute (T.E.I.) of Kalamata, Greece University of Patras, Greece University of Western Greece, Greece
X
Organization
Adam Adamopoulos Athanasios Tsakalidis Efthyvoulos Kyriacou Elias Maglogiannis George Anastassopoulos
University of Thrace, Greece University of Patras, Greece Frederick University, Cyprus University of Central Greece, Greece Democritus University of Thrace, Greece
Honorary Chairs Tharam Dillon Max Bramer
Curtin University, Australia University of Portsmouth, UK
Referees Aldanondo M. Alexandridis G. Anagnostou K. Anastassopoulos G. Andreadis I. Andreou A. Avlonitis M. Bankovic Z. Bessis N. Boracchi G. Caridakis George Charalambous C. Chatzioannou Aristotle Constantinides A. Damoulas T. Doukas Charalampos Fox C. Gaggero M. Gammerman Alex Georgopoulos E. Hatzilygeroudis I. Hunt S. Janssens G. Kabzinski J. Kameas A.
Karpouzis Kostas Kefalas P. Kermanidis I. Kosmopoulos Dimitrios Kosmopoulos D. Kyriacou E. Lazaro J.Lopez Likothanassis S. Lorentzos Nikos Malcangi M. Maragkoudakis M. Marcelloni F. Margaritis K. Mohammadian M. Nouretdinov Ilia Olej Vladimir Onaindia E. Papatheocharous E. Plagianakos Vassilis Portinale L. Rao Vijan Roveri M. Ruggero Donida Labati Sakelariou I. Schizas C.
Senatore S. Shen Furao Sideridis A. Sioutas S. Sotiropoulos D.G. Stafylopatis A. Tsadiras A. Tsakalidis Athanasios Tsapatsoulis N. Tsevas S. Vassileiades N. Verykios V. Vishwanathan Mohan Voulgaris Z. Vouros G. Vouyioukas Demosthenis Vovk Volodya Wallace Manolis Wang Zidong Wyns B. Xrysikopoulos V. Yialouris K. Zenker Bernd Ludwig Zhiyuan Luo
Sponsoring Institutions The 12th EANN / 7th AIAI Joint Conferences were organized by IFIP (International Federation for Information Processing), INNS (International Neural Network Society), the Aristotle University of Thessaloniki, the Democritus University of Thrace and the Ionian University of Corfu.
Table of Contents – Part II
Computer Vision and Robotics Real Time Robot Policy Adaptation Based on Intelligent Algorithms . . . Genci Capi, Hideki Toda, and Shin-Ichiro Kaneko
1
A Model and Simulation of Early-Stage Vision as a Developmental Sensorimotor Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Olivier L. Georgeon, Mark A. Cohen, and Am´elie V. Cordier
11
Enhanced Object Recognition in Cortex-Like Machine Vision . . . . . . . . . . Aristeidis Tsitiridis, Peter W.T. Yuen, Izzati Ibrahim, Umar Soori, Tong Chen, Kan Hong, Zhengjie Wang, David James, and Mark Richardson
17
Classification - Pattern Recognition A New Discernibility Metric and Its Application on Pattern Classification and Feature Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zacharias Voulgaris
27
Financial and Management Applications of AI Time Variations of Association Rules in Market Basket Analysis . . . . . . . Vasileios Papavasileiou and Athanasios Tsadiras A Software Platform for Evolutionary Computation with Pluggable Parallelism and Quality Assurance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pedro Evangelista, Jorge Pinho, Emanuel Gon¸calves, Paulo Maia, Jo˜ ao Luis Sobral, and Miguel Rocha Financial Assessment of London Plan Policy 4A.2 by Probabilistic Inference and Influence Diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Amin Hosseinian-Far, Elias Pimenidis, Hamid Jahankhani, and D.C. Wijeyesekera Disruption Management Optimization for Military Logistics . . . . . . . . . . . Ayda Kaddoussi, Nesrine Zoghlami, Hayfa Zgaya, Slim Hammadi, and Francis Bretaudeau
36
45
51
61
XII
Table of Contents – Part II
Fuzzy Systems Using a Combined Intuitionistic Fuzzy Set-TOPSIS Method for Evaluating Project and Portfolio Management Information Systems . . . . Vassilis C. Gerogiannis, Panos Fitsilis, and Achilles D. Kameas
67
Fuzzy and Neuro-Symbolic Approaches to Assessment of Bank Loan Applicants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ioannis Hatzilygeroudis and Jim Prentzas
82
Comparison of Fuzzy Operators for IF-Inference Systems of Takagi-Sugeno Type in Ozone Prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vladim´ır Olej and Petr H´ ajek
92
LQR-Mapped Fuzzy Controller Applied to Attitude Stabilization of a Power-Aided-Unicycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ping-Ho Chen, Wei-Hsiu Hsu, and Ding-Shinan Fong
98
Optimal Fuzzy Controller Mapped from LQR under Power and Torque Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ping-Ho Chen and Kuang-Yow Lian
104
Learning and Novel Algorithms A New Criterion for Clusters Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hosein Alizadeh, Behrouz Minaei, and Hamid Parvin
110
Modeling and Dynamic Analysis on Animals’ Repeated Learning Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mu Lin, Jinqiao Yang, and Bin Xu
116
Generalized Bayesian Pursuit: A Novel Scheme for Multi-Armed Bernoulli Bandit Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xuan Zhang, B. John Oommen, and Ole-Christoffer Granmo
122
Recurrent and Radial Basis Function ANN A Multivalued Recurrent Neural Network for the Quadratic Assignment Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Graci´ an Trivi˜ no, Jos´e Mu˜ noz, and Enrique Dom´ınguez Employing a Radial-Basis Function Artificial Neural Network to Classify Western and Transition European Economies Based on the Emissions of Air Pollutants and on Their Income . . . . . . . . . . . . . . . . . . . . Kyriaki Kitikidou and Lazaros Iliadis
132
141
Table of Contents – Part II
XIII
Machine Learning Elicitation of User Preferences via Incremental Learning in a Declarative Modelling Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Georgios Bardis, Vassilios Golfinopoulos, Dimitrios Makris, Georgios Miaoulis, and Dimitri Plemenos Predicting Postgraduate Students’ Performance Using Machine Learning Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Maria Koutina and Katia Lida Kermanidis
150
159
Generic Algorithms Intelligent Software Project Scheduling and Team Staffing with Genetic Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Constantinos Stylianou and Andreas S. Andreou
169
Data Mining Comparative Analysis of Content-Based and Context-Based Similarity on Musical Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Boletsis, A. Gratsani, D. Chasanidou, I. Karydis, and K. Kermanidis
179
Learning Shallow Syntactic Dependencies from Imbalanced Datasets: A Case Study in Modern Greek and English . . . . . . . . . . . . . . . . . . . . . . . . . Argiro Karozou and Katia Lida Kermanidis
190
A Random Forests Text Transliteration System for Greek Digraphia . . . . Alexandros Panteli and Manolis Maragoudakis
196
Acceptability in Timed Frameworks with Intermittent Arguments . . . . . . Maria Laura Cobo, Diego C. Martinez, and Guillermo R. Simari
202
Object Oriented Modelling in Information Systems Based on Related Text Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kolyo Onkov
212
Reinforcement Learning Ranking Functions in Large State Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . Klaus H¨ aming and Gabriele Peters
219
Web Applications of ANN Modelling of Web Domain Visits by Radial Basis Function Neural Networks and Support Vector Machine Regression . . . . . . . . . . . . . . . . . . . Vladim´ır Olej and Jana Filipov´ a
229
XIV
Table of Contents – Part II
A Framework for Web Page Rank Prediction . . . . . . . . . . . . . . . . . . . . . . . . Elli Voudigari, John Pavlopoulos, and Michalis Vazirgiannis
240
Towards a Semantic Calibration of Lexical Word via EEG . . . . . . . . . . . . . Marios Poulos
250
Medical Applications of ANN and Ethics of AI Data Mining Tools Used in Deep Brain Stimulation – Analysis Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Oana Geman
259
Reliable Probabilistic Prediction for Medical Decision Support . . . . . . . . . Harris Papadopoulos
265
Cascaded Window Memoization for Medical Imaging . . . . . . . . . . . . . . . . . Farzad Khalvati, Mehdi Kianpour, and Hamid R. Tizhoosh
275
Fast Background Elimination in Fluorescence Microbiology Images: Comparison of Four Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shan Gong and Antonio Art´es-Rodr´ıguez
285
Experimental Verification of the Effectiveness of Mammography Testing Description’s Standardization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Teresa Podsiadly-Marczykowska and Rafal Zawi´slak
291
Ethical Issues of Artificial Biomedical Applications . . . . . . . . . . . . . . . . . . . Athanasios Alexiou, Maria Psixa, and Panagiotis Vlamos
297
Environmental and Earth Applications of AI ECOTRUCK: An Agent System for Paper Recycling . . . . . . . . . . . . . . . . . Nikolaos Bezirgiannis and Ilias Sakellariou Prediction of CO and NOx levels in Mexico City Using Associative Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Amadeo Arg¨ uelles, Cornelio Y´ an ˜ez, Itzam´ a L´ opez, and Oscar Camacho Neural Network Approach to Water-Stressed Crops Detection Using Multispectral WorldView-2 Satellite Imagery . . . . . . . . . . . . . . . . . . . . . . . . ´ Dubravko Culibrk, Predrag Lugonja, Vladan Mini´c, and Vladimir Crnojevi´c A Generalized Fuzzy-Rough Set Application for Forest Fire Risk Estimation Feature Reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . T. Tsataltzinos, L. Iliadis, and S. Spartalis
303
313
323
332
Table of Contents – Part II
Pollen Classification Based on Geometrical, Descriptors and Colour Features Using Decorrelation Stretching Method . . . . . . . . . . . . . . . . . . . . . Jaime R. Ticay-Rivas, Marcos del Pozo-Ba˜ nos, Carlos M. Travieso, Jorge Arroyo-Hern´ andez, Santiago T. P´erez, Jes´ us B. Alonso, and Federico Mora-Mora
XV
342
Computational Intelligence in Software Engineering (CISE) Workshop Global Optimization of Analogy-Based Software Cost Estimation with Genetic Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dimitrios Milios, Ioannis Stamelos, and Christos Chatzibagias The Impact of Sampling and Rule Set Size on Generated Fuzzy Inference System Predictive Accuracy: Analysis of a Software Engineering Data Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stephen G. MacDonell
350
360
Intelligent Risk Identification and Analysis in IT Network Systems . . . . . Masoud Mohammadian
370
Benchmark Generator for Software Testers . . . . . . . . . . . . . . . . . . . . . . . . . . Javier Ferrer, Francisco Chicano, and Enrique Alba
378
Automated Classification of Medical-Billing Data . . . . . . . . . . . . . . . . . . . . R. Crandall, K.J. Lynagh, T. Mehoke, and N. Pepper
389
Artificial Intelligence Applications in Biomedicine (AIAB) Workshop Brain White Matter Lesions Classification in Multiple Sclerosis Subjects for the Prognosis of Future Disability . . . . . . . . . . . . . . . . . . . . . . . Christos P. Loizou, Efthyvoulos C. Kyriacou, Ioannis Seimenis, Marios Pantziaris, Christodoulos Christodoulou, and Constantinos S. Pattichis
400
Using Argumentation for Ambient Assisted Living . . . . . . . . . . . . . . . . . . . Julien Marcais, Nikolaos Spanoudakis, and Pavlos Moraitis
410
Modelling Nonlinear Responses of Resonance Sensors in Pressure Garment Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Timo Salpavaara and Pekka Kumpulainen
420
An Adaptable Framework for Integrating and Querying Sensor Data . . . Shahina Ferdous, Sarantos Kapidakis, Leonidas Fegaras, and Fillia Makedon
430
XVI
Table of Contents – Part II
Feature Selection by Conformal Predictor . . . . . . . . . . . . . . . . . . . . . . . . . . . Meng Yang, Ilia Nouretdunov, Zhiyuan Luo, and Alex Gammerman
439
Applying Conformal Prediction to the Bovine TB Diagnosing . . . . . . . . . . Dmitry Adamskiy, Ilia Nouretdinov, Andy Mitchell, Nick Coldham, and Alex Gammerman
449
Classifying Ductal Tree Structures Using Topological Descriptors of Branching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Angeliki Skoura, Vasileios Megalooikonomou, Predrag R. Bakic, and Andrew D.A. Maidment Intelligent Selection of Human miRNAs and Mouse mRNAs Related to Obstructive Nephropathy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ioannis Valavanis, P. Moulos, Ilias Maglogiannis, Julie Klein, Joost Schanstra, and Aristotelis Chatziioannou Independent Component Clustering for Skin Lesions Characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S.K. Tasoulis, C.N. Doukas, I. Maglogiannis, and V.P. Plagianakos A Comparison of Venn Machine with Platt’s Method in Probabilistic Outputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chenzhe Zhou, Ilia Nouretdinov, Zhiyuan Luo, Dmitry Adamskiy, Luke Randell, Nick Coldham, and Alex Gammerman Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
455
464
472
483
491
Table of Contents – Part I
Computer Vision and Robotics ART-Based Fusion of Multi-modal Information for Mobile Robots . . . . . . Elmar Bergh¨ ofer, Denis Schulze, Marko Tscherepanow, and Sven Wachsmuth Vision-Based Autonomous Navigation Using Supervised Learning Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jefferson R. Souza, Gustavo Pessin, Fernando S. Os´ orio, and Denis F. Wolf
1
11
Self Organizing Maps SOM-Based Clustering and Optimization of Production . . . . . . . . . . . . . . . Primoˇz Potoˇcnik, Tomaˇz Berlec, Marko Starbek, and Edvard Govekar Behavioral Profiles for Building Energy Performance Using eXclusive SOM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F´elix Iglesias V´ azquez, Sergio Cantos Gaceo, Wolfgang Kastner, and Jos´e A. Montero Morales
21
31
Classification - Pattern Recognition Hypercube Neural Network Algorithm for Classification . . . . . . . . . . . . . . . Dominic Palmer-Brown and Chrisina Jayne
41
Improving the Classification Performance of Liquid State Machines Based on the Separation Property . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Emmanouil Hourdakis and Panos Trahanias
52
A Scale-Changeable Image Analysis Method . . . . . . . . . . . . . . . . . . . . . . . . . Hui Wei, Bo Lang, and Qing-song Zuo
63
Induction of Linear Separability through the Ranked Layers of Binary Classifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Leon Bobrowski
69
Classifying the Differences in Gaze Patterns of Alphabetic and Logographic L1 Readers - A Neural Network Approach . . . . . . . . . . . . . . . Andr´e Frank Krause, Kai Essig, Li-Ying Essig-Shih, and Thomas Schack
78
XVIII
Table of Contents – Part I
Subspace-Based Face Recognition on an FPGA . . . . . . . . . . . . . . . . . . . . . . Pablo Pizarro and Miguel Figueroa A Window-Based Self-Organizing Feature Map (SOFM) for Vector Filtering Segmentation of Color Medical Imagery . . . . . . . . . . . . . . . . . . . . Ioannis M. Stephanakis, George C. Anastassopoulos, and Lazaros Iliadis
84
90
Financial and Management Applications of AI Neural Network Rule Extraction to Detect Credit Card Fraud . . . . . . . . . Nick F. Ryman-Tubb and Paul Krause
101
Fuzzy Systems A Neuro-Fuzzy Hybridization Approach to Model Weather Operations in a Virtual Warfare Analysis System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Vijay Rao, Lazaros Iliadis, and Stefanos Spartalis
111
Employing Smart Logic to Spot Audio in Real Time on Deeply Embedded Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mario Malcangi
122
Support Vector Machines Quantization of Adulteration Ratio of Raw Cow Milk by Least Squares Support Vector Machines (LS-SVM) and Visible/Near Infrared Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ching-Lu Hsieh, Chao-Yung Hung, and Ching-Yun Kuo Support Vector Machines versus Artificial Neural Networks for Wood Dielectric Loss Factor Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lazaros Iliadis, Stavros Tachos, Stavros Avramidis, and Shawn Mansfield
130
140
Learning and Novel Algorithms Time-Frequency Analysis of Hot Rolling Using Manifold Learning . . . . . . ´ Francisco J. Garc´ıa, Ignacio D´ıaz, Ignacio Alvarez, Daniel P´erez, Daniel G. Ordonez, and Manuel Dom´ınguez
150
Reinforcement and Radial Basis Function ANN Application of Radial Bases Function Network and Response Surface Method to Quantify Compositions of Raw Goat Milk with Visible/Near Infrared Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ching-Lu Hsieh, Chao-Yung Hung, and Mei-Jen Lin
156
Table of Contents – Part I
Transferring Models in Hybrid Reinforcement Learning Agents . . . . . . . . Anestis Fachantidis, Ioannis Partalas, Grigorios Tsoumakas, and Ioannis Vlahavas
XIX
162
Machine Learning Anomaly Detection from Network Logs Using Diffusion Maps . . . . . . . . . . Tuomo Sipola, Antti Juvonen, and Joel Lehtonen Large Datasets: A Mixed Method to Adapt and Improve Their Learning by Neural Networks Used in Regression Contexts . . . . . . . . . . . . . . . . . . . . Marc Sauget, Julien Henriet, Michel Salomon, and Sylvain Contassot-Vivier
172
182
Evolutionary Genetic Algorithms - Optimization Evolutionary Algorithm Optimization of Edge Delivery Sites in Next Generation Multi-service Content Distribution Networks . . . . . . . . . . . . . . Ioannis Stephanakis and Dimitrios Logothetis Application of Neural Networks to Morphological Assessment in Bovine Livestock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Horacio M. Gonz´ alez-Velasco, Carlos J. Garc´ıa-Orellana, Miguel Mac´ıas-Mac´ıas, Ram´ on Gallardo-Caballero, and Antonio Garc´ıa-Manso
192
203
Web Applications of ANN Object Segmentation Using Multiple Neural Networks for Commercial Offers Visual Search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I. Gallo, A. Nodari, and M. Vanetti
209
Spiking ANN Method for Training a Spiking Neuron to Associate Input-Output Spike Trains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ammar Mohemmed, Stefan Schliebs, Satoshi Matsuda, and Nikola Kasabov
219
Feature Extraction - Minimization Two Different Approaches of Feature Extraction for Classifying the EEG Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pari Jahankhani, Juan A. Lara, Aurora P´erez, and Juan P. Valente
229
XX
Table of Contents – Part I
An Ensemble Based Approach for Feature Selection . . . . . . . . . . . . . . . . . . Behrouz Minaei-Bidgoli, Maryam Asadi, and Hamid Parvin A New Feature Extraction Method Based on Clustering for Face Recognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sabra El Ferchichi, Salah Zidi, Kaouther Laabidi, Moufida Ksouri, and Salah Maouche
240
247
Medical Applications of AI A Recurrent Neural Network Approach for Predicting Glucose Concentration in Type-1 Diabetic Patients . . . . . . . . . . . . . . . . . . . . . . . . . . Fayrouz Allam, Zaki Nossai, Hesham Gomma, Ibrahim Ibrahim, and Mona Abdelsalam Segmentation of Breast Ultrasound Images Using Neural Networks . . . . . Ahmed A. Othman and Hamid R. Tizhoosh Knowledge Discovery and Risk Prediction for Chronic Diseases: An Integrated Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anju Verma, Maurizio Fiasch´e, Maria Cuzzola, Francesco C. Morabito, and Giuseppe Irrera Permutation Entropy for Discriminating ‘Conscious’ and ‘Unconscious’ State in General Anesthesia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nicoletta Nicolaou, Saverios Houris, Pandelitsa Alexandrou, and Julius Georgiou
254
260
270
280
Environmental and Earth Applications of AI Determining Soil – Water Content by Data Driven Modeling When Relatively Small Data Sets Are Available . . . . . . . . . . . . . . . . . . . . . . . . . . . Milan Cisty
289
A Neural Based Approach and Probability Density Approximation for Fault Detection and Isolation in Nonlinear Systems . . . . . . . . . . . . . . . . . . . P. Boi and A. Montisci
296
A Neural Network Tool for the Interpolation of foF2 Data in the Presence of Sporadic E Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Haris Haralambous, Antonis Ioannou, and Harris Papadopoulos
306
Multi Layer ANN Neural Networks Approach to Optimization of Steel Alloys Composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Petia Koprinkova-Hristova, Nikolay Tontchev, and Silviya Popova
315
Table of Contents – Part I
XXI
Predictive Automated Negotiators Employing Risk-Seeking and Risk-Averse Strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Marisa Masvoula, Constantin Halatsis, and Drakoulis Martakos
325
Maximum Shear Modulus Prediction by Marchetti Dilatometer Test Using Neural Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Manuel Cruz, Jorge M. Santos, and Nuno Cruz
335
NNIGnets, Neural Networks Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tˆ ania Fontes, Vˆ ania Lopes, Lu´ıs M. Silva, Jorge M. Santos, and Joaquim Marques de S´ a Key Learnings from Twenty Years of Neural Network Applications in the Chemical Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Aaron J. Owens
345
351
Bioinformatics Incremental- Adaptive- Knowledge Based- Learning for Informative Rules Extraction in Classification Analysis of aGvHD . . . . . . . . . . . . . . . . . Maurizio Fiasch´e, Anju Verma, Maria Cuzzola, Francesco C. Morabito, and Giuseppe Irrera
361
The Applications of Soft Computing to Telecommunications (ASCOTE) Workshop An Intelligent Approach to Detect Probe Request Attacks in IEEE 802.11 Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Deepthi N. Ratnayake, Hassan B. Kazemian, Syed A. Yusuf, and Azween B. Abdullah An Intelligent Keyboard Framework for Improving Disabled People Computer Accessibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Karim Ouazzane, Jun Li, and Hassan B. Kazemian Finding 3G Mobile Network Cells with Similar Radio Interface Quality Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pekka Kumpulainen, Mika S¨ arkioja, Mikko Kylv¨ aj¨ a, and Kimmo H¨ at¨ onen Analyzing 3G Quality Distribution Data with Fuzzy Rules and Fuzzy Clustering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pekka Kumpulainen, Mika S¨ arkioja, Mikko Kylv¨ aj¨ a, and Kimmo H¨ at¨ onen Adaptive Service Composition for Meta-searching in a Mobile Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ronnie Cheung and Hassan B. Kazemian
372
382
392
402
412
XXII
Table of Contents – Part I
Simulation of Web Data Traffic Patterns Using Fractal Statistical Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shanyu Tang and Hassan B. Kazemian
422
Computational Intelligence Applications in Bioinformatics (CIAB) Workshop Mathematical Models of Dynamic Behavior of Individual Neural Networks of Central Nervous System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dimitra-Despoina Pagania, Adam Adamopoulos, and Spiridon D. Likothanassis
433
Towards Optimal Microarray Universal Reference Sample Designs: An In-Silico Optimization Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . George Potamias, Sofia Kaforou, and Dimitris Kafetzopoulos
443
Information-Preserving Techniques Improve Chemosensitivity Prediction of Tumours Based on Expression Profiles . . . . . . . . . . . . . . . . . . E.G. Christodoulou, O.D. Røe, A. Folarin, and I. Tsamardinos
453
Optimizing Filter Processes on Protein Interaction Clustering Results Using Genetic Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Charalampos Moschopoulos, Grigorios Beligiannis, Sophia Kossida, and Spiridon Likothanassis Adaptive Filtering Techniques Combined with Natural Selection-Based Heuristic Algorithms in the Prediction of Protein-Protein Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Christos M. Dimitrakopoulos, Konstantinos A. Theofilatos, Efstratios F. Georgopoulos, Spyridon D. Likothanassis, Athanasios K. Tsakalidis, and Seferina P. Mavroudi
463
471
Informatics and Intelligent Systems Applications for Quality of Life information Services (ISQLIS) Workshop Investigation of Medication Dosage Influences from Biological Weather . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kostas Karatzas, Marina Riga, Dimitris Voukantsis, and ˚ Asl¨ og Dahl Combination of Survival Analysis and Neural Networks to Relate Life Expectancy at Birth to Lifestyle, Environment, and Health Care Resources Indicators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lazaros Iliadis and Kyriaki Kitikidou
481
491
Table of Contents – Part I
An Artificial Intelligence-Based Environment Quality Analysis System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mihaela Oprea and Lazaros Iliadis Personalized Information Services for Quality of Life: The Case of Airborne Pollen Induced Symptoms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dimitris Voukantsis, Kostas Karatzas, Siegfried Jaeger, and Uwe Berger Fuzzy Modeling of the Climate Change Effect to Drought and to Wild Fires in Cyprus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xanthos Papakonstantinou, Lazaros S. Iliadis, Elias Pimenidis, and Fotis Maris Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
XXIII
499
509
516
529
Real Time Robot Policy Adaptation Based on Intelligent Algorithms Genci Capi1, Hideki Toda1, and Shin-Ichiro Kaneko2 1
Department of Electric and Electronic Eng., University of Toyama, Toyama, Japan
[email protected] 2 Toyama National College of Technology
Abstract. In this paper we present a new method for robot real time policy adaptation by combining learning and evolution. The robot adapts the policy as the environment conditions change. In our method, we apply evolutionary computation to find the optimal relation between reinforcement learning parameters and robot performance. The proposed algorithm is evaluated in the simulated environment of the Cyber Rodent (CR) robot, where the robot has to increase its energy level by capturing the active battery packs. The CR robot lives in two environments with different settings that replace each other four times. Results show that evolution can generate an optimal relation between the robot performance and exploration-exploitation of reinforcement learning, enabling the robot to adapt online its strategy as the environment conditions change. Keywords: Reinforcement computation.
learning,
policy
adaptation,
evolutionary
1 Introduction Reinforcement learning (RL) ([1], [2]) is an efficient learning framework for autonomous robots, in which the robot learns how to behave, from interactions with the environment, without explicit environment models or teacher signals. Most RL applications, so far, have been constrained to stationary environments. However, in many real-world tasks, the environment is not fixed. Therefore, the robot must change its strategy based on the environment conditions. For small environment changes, Minato et al., (2000) has pointed out that current knowledge learned in a previous environment is partially applicable even after the environment has changed, if we only consider reaching the goal and thereby sacrifice optimality ([3]). Efforts have also been made to move in more dynamic environments. Matsui et al. ([4]) proposed a method, which senses a changing environment by collecting failed instances and partially modifies the strategy for adapting to subsequent changes of the environment by reinforcement learning. Doya incorporated a noise term in policies, in order to promote exploration ([5]). The size of noise is reduced as the performance improves. However, this method can be applied when the value function is known for all the states. L. Iliadis et al. (Eds.): EANN/AIAI 2011, Part II, IFIP AICT 364, pp. 1–10, 2011. © IFIP International Federation for Information Processing 2011
2
G. Capi, H. Toda, and S.-I. Kaneko
Previous approaches on combining learning and evolution ([6], [7], [8], [9]) reported that combination tends to provide earlier achievement of superior performance. Niv et al. have considered evolution of RL in uncertain environments ([10]). They solve near-optimal neuronal learning rules in order to allow simulated bees to respond rapidly to changes in reward contingencies. In our previous work, we considered evolution of metaparameters for faster convergence of reinforcement learning ([11], [12]). However, in all these approaches the robot learned the optimal strategy in stationary environment. In difference from previous works, in this paper, we combine an actor-critic RL and evolution to develop robots able to adapt their strategy as the environment changes. The metaparameters, initial weight connection, number of hidden neurons of actor and critic networks, and the relation between the energy level and cooling factor are evolved by a real number Genetic Algorithm (GA). In order to test the effectiveness of the proposed algorithm, we considered a biologically inspired task for the CR robot ([13]). The robot must survive and increase its energy level by capturing the active battery packs distributed in the environment. The robot lives in two different environments, which substitute for each other four times during the robot’s life. Therefore, the robot must adapt its strategy as the environment changes. The performance of proposed method is compared with that of (a) RL and (b) evolution of neural controller. In the actor-critic RL, we used arbitrarily selected metaparameters, randomly initialized initial weight connections, and a linearly proportional relationship between cooling factor and energy level. The robot controlled by the evolved neural controller applies the same strategy throughout all its life, which was optimal only for one environment. The performance of the actor-critic RL was strongly related to the metaparameters, especially the relation between cooling factor and energy level. Combining learning and evolution gives the best performance overall. Because of optimized metaparameters and initial weight connections, the robot was able to exploit the environment from the beginning of its life. In addition, the robot switched between exploration and exploitation based on the optimized relation between the energy level and cooling factor.
2 Cyber Rodent Robot In our simulations, we used the CR robot, which is a two-wheel-driven mobile robot, as shown in Fig. 1. The CR is 250 mm long and weights 1.7 kg. The CR is equipped with: • • • • • • • • •
Omni-directional C-MOS camera. IR range sensor. Seven IR proximity sensors. 3-axis acceleration sensor. 2-axis gyro sensor. Red, green and blue LED for visual signaling. Audio speaker and two microphones for acoustic communication. Infrared port to communicate with a nearby robot. Wireless LAN card and USB port to communicate with the host computer.
Real Time Robot Policy Adaptation Based on Intelligent Algorithms
3
Five proximity sensors are positioned on the front of robot, one behind and one under the robot pointing downwards. The proximity sensor under the robot is used when the robot moves wheelie. The CR contains a Hitachi SH-4 CPU with 32 MB memory. The FPGA graphic processor is used for video capture and image processing at 30 Hz. 2.1 Environment The CR robot has to survive and increase its energy level by capturing the active battery packs distributed in a rectangular environment of 2.5m x 3.5m (Fig. 2). The active battery packs have a red LED. After the charging time, the battery pack becomes inactive and its LED color changes to green and the battery becomes active again after the reactivation time. The CR robot is initially placed in a random position and orientation. The robot lives in two different environments that alternatively substitute eachother four times. Based on environments settings, the robot must learn different policies in order to survive and increase its energy level. As shown in Fig. 2, the first and second environments have eight and two battery packs, respectively. In the first environment, the batteries have a long reactivation time. In addition, the energy consumed for 1m motion is low. Therefore, the best policy is to capture any visible battery pack (the nearest when there are more than one). When there is no visible active battery pack, the robot have to search in the environment. In the second environment, the reactivation time is short and the energy consumed during robot motion is increased. Therefore, the optimal policy is to wait until the previously captured battery pack becomes active again rather than searching for other active battery packs.
Fig. 1. CR robot and the battery pack
(a) Environment 1
(b) Environment 2
Fig. 2. Environments
4
G. Capi, H. Toda, and S.-I. Kaneko
3 Intelligent Algorithms Consider the Cyber Rodent robot in an environment where at any given time t, the robot is able to choose an action. Also, at any given time t, the environment provides the robot with a reward rt. Our implementation of the actor-critic has three parts: 1) an input layer of robot state; 2) a critic network that learns appropriate weights from the state to enable it to output information about the value of particular state; 3) an actor network that learns the appropriate weights from the state, which enable it to represent the action the robot should make in a particular state. Each time step, the robot selects one of the following actions: 1) Capture the battery pack; 2) Search for a battery pack; 3) Wait for a determined period of time. The wait behavior is interrupted if a battery becomes active. Both networks receive as input a constant bias input, the battery level and distance to the nearest active battery pack (both normalized between 0 and 1). 3.1 Critic The standard approach is for the critic to attempt to learn the value function, V(x), which is really an evaluation of the actions currently specified by the actor. The value function is usually defined as, for any state x, the discounted total future reward that is expected, on average, to accrue after being in state x and then following the actions currently specified by the actor. If xt is the state at time t, we may define the value as: V(xt)=