Lecture Notes in Artificial Intelligence

Transcription

1 Lecture Notes in Artificial Intelligence 1152 Subseries of Lecture Notes in Computer Science Edited by J. G. Carbonell and J. Siekmann Lecture Notes in Computer Science Edited by G. Goos, J. Hartmanis and J. van Leeuwen

2 Takeshi Furuhashi Yoshiki Uchikawa (Eds.) Fuzzy Logic, Neural Networks, and Evolutionary Computation IEEE/Nagoya-University World Wisepersons Workshop Nagoya, Japan, November 14-15, 1995 Selected Papers Springer

3 Series Editors Jaime G. Carbonell, Carnegie Mellon University, Pittsburgh, PA, USA J6rg Siekmann, University of Saarland, Saarbrticken, Germany Volume Editors Takeshi Furuhashi Yoshiki Uchikawa Nagoya University, Department of Information Electronics Furo-cho, Chikusa-ku, Nagoya , Japan { nuee.nagoya-u.ac.jp Cataloging-in-Publication Data applied for Die Deutsche Bibliothek - CIP-Einheitsaufnahme Fuzzy logic, neural networks and evolutionary computation : selected papers / IEEE/Nagoya-University World Wisepersons Workshop, Nagoya, Japan, November 14-15, Takeshi Furuhashi ; Yoshiki Uchikawa (ed.). - Berlin ; Heidelberg ; New York ; Barcelona ; Budapest ; Hong Kong ; London ; Milan ; Paris ; Santa Clara ; Singapore ; Tokyo : Springer, 1996 (Lecture notes in computer science ; Vol : Lecture notes in artificial intelligence) ISBN NE: Furuhashi, Takeshi [Hrsg.]; World Wisepersons Workshop <4, 1995, Nagoya>; Industrial Electronics Society; GT CR Subject Classification (1991): 1.2,F.1-2 ISBN Springer-Verlag Berlin Heidelberg New York 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 its current version, and permission for use must always be obtained from Springer -Verlag. Violations are liable for prosecution under the German Copyright Law. 9 Springer-Vertag Berlin Heidelberg 1996 Printed in Germany Typesetting: Camera ready by author SPIN / Printed on acid-free paper

4 Preface Fuzzy logic is one of the key technologies for representing human knowledge in the brain and for constructing adaptive systems. Difficulties for fuzzy logic include knowledge acquisition from experts and/or knowledge finding for unknown tasks. Neural networks (NNs) and evolutionary computation (EC) are also gaining attention for their potential for knowledge representation/finding. Combining the technologies of fuzzy logic and neural networks/evolutionary computation is expected to open a new paradigm in machine learning for the realization of human-like information generating system. This is the background against which the IEEE/Nagoya University World Wisepersons Workshop (WWW'95) on Fuzzy Logic and Neural Networks/ Evolutionary Computation was organized. This workshop is the fourth of the WWW series organized to cover the rapidly advancing field of research on the combined technologies of Fuzzy - NN and Fuzzy - EC. IEEE/Nagoya University WWW is a series of workshops sponsored by Nagoya University and co-sponsored by the IEEE Industrial Electronics Society. The third one (WWW'94) was organized by Associate Prof. Furuhashi (Dept. of Information Electronics, Nagoya University) in August 1994 to allow attendees to have a lively discussion on the subject "Present and Future Situation of the Combination Technologies of Fuzzy Logic and Neural Networks/Genetic Algorithm", an area which has made remarkable progress recently. The selected papers of WWW'94 were published as LNAI/LNCS Vol in The fourth WWW is to extend and deepen the discussions from WWW'94. The steering committee has accepted 17 excellent papers. As the editors for the selected papers of WWW'95, we selected twelve excellent papers which were revised, extended, and rigorously reviewed by three reviewers each. The first five papers discuss interesting and advancing topic on the combined technology Fuzzy - EC. M. A. Lee, H. Esbensen, and L. Lemaitre develop techniques to design mutiobjective optimization algorithms based on hybrid fuzzy system/evolutionary algorithm techniques. The technique is based on pioneering work done by one of the authors which uses a fuzzy system to control the evolutionary algorithm. S. Nakanishi et al. propose a new fuzzy modeling method using the genetic algorithm. Structure identification of the fuzzy models has been one of the difficult tasks of the fuzzy modeling. Hierarchical structures of unknown objects can be identified from the input and output data by the proposed method. H. Ishibuchi, T. Nakashima, and T. Murata present a fuzzy classifier system for automatic generation of a compact fuzzy system based on a small number of linguistic classification rules. T. Nomura and T. Miyoshi propose a new genetic operator called Unfair Average Crossover for fuzzy clustering using real value coding. The parameters of membership

5 vi functions of fuzzy if-then rules are directly encoded. T. Yoshikawa et al. present a new coding method using artificial DNA based on an analogy with biological DNA. A new mechanism of development from the artificial DNA is also presented. Redundancy and overlaps of genes are the unique features of the proposed method. Fuzzy rules are found by the proposed combination technology of the new method. In the sixth paper, T. Aoki et al. propose a new architecture consisting of hierarchical fuzzy rules, a fuzzy performance evaluation system, and learning automata. The evaluation function of the learning automata is constructed using fuzzy logic for coping with a dynamically changing environment. The next three papers discuss hybrid technology of Fuzzy - NN. The seventh paper discusses a timely and important problem of fuzzy modeling. P. J. Costa Branco, N. Lori, and J. A. Dente study two pre-processing methods for structure identification of fuzzy models: a statistical method called Principal Component Analysis and a clustering technique called Autonomous Mountain-Clustering Method. These two methods are effective for efficient fuzzy modeling, with neuro-fuzzy techniques. The eighth paper, by A. Z. Kouzani and A. Bouzerdoum, introduces a generic fuzzy neuron. A fuzzy neural network architecture based on the generic fuzzy neuron is developed for motion estimation. A. Buller presents a unique Fizzy-Fuzzy inference method utilizing a working memory inhabited by a society of Running Agents (RATs). This method is devised for fuzzy knowledge processing in a massively parallel way. The last three papers also tackle interesting problems. The tenth paper, presented by E. Perez-Minana, P. Ross and J Hallam, describes multi-layer perceptron design utilizing the information subsumed in the Delaunay Triangulation and Voronoi Diagram for achieving a reasonable network for the task of interest. M. Inuiguchi et al. successfully applied a genetic algorithm to an actual problem: coal purchase planning in a real electric power plant. The last paper, by H. de Garis, reports a challenging project, the CAM-Brain Project, which aims to use evolutionary engineering techniques to build/grow/evolve a RAM-and-cellular-automata based artificial brain consisting of thousands of interconnected neural network modules inside special hardware. I am now proud to have these papers published in the LNCS/LNAI series. Nagoya, July 1996 TakeshiFuruhashi YoshikiUchikawa

6 Contents 1. Fuzzy & Evolutionary Computation The Design of tlybrid Fuzzy~Evolutionary Optimization Algorithms Michael A. Lee, Henrik Esbensen, Laurent Lemaitre Multiobjective Structure Identification of Acquired Knowledge in Fuzzy Inference by Genetic Algorithms Shohachiro Nakanishi, Akihiro Ohtake, Ronald R. Yager, Shinobu Ohtani, Hiroaki Kikuchi A Fuzzy Classifier System That Generates Linguistic Rules for Pattern Classification Problems Hisao Ishibuchi, Tomoharu Nakashima, Tadahiko Murata Numerical Coding and Unfair Average Crossover in GA for Fuzzy Rule Extraction in Dynamic Environments Tatsuya Nomura, Tsutomu Miyoshi Acquisition of Fuzzy Rules from DNA Coding Method Tomohiro Yoshikawa, Takeshi Furuhashi, Yoshiki Uchikawa Fuzzy & Learning Automata Experimental Study on Acquisition of Optimal Action for Autonomous Mobile Robot to Avoid Moving Multiobstacles Takeshi Aoki, Toshiaki Oka, Soichiro Hayakawa, Tatsuya Suzuki, Shigeru Okuma Fuzzy & Neural Networks New Approaches on Structure Identification of Fuzzy Models: Case Study in an Electro-Mechanical System P.J. Costa Branco, N. Lori, J.A. Dente 104

7 viii A Generic Fuzzy Neuron and Its Application to Motion Estimation Abbas Z. Kouzani, Abdesselam Bouzerdoum Fizzy-Fuzzy Inferencing Andrzej Buller NN, GA, CAM-Brain Multi-layer Perceptron Design Using Delaunay Triangulations Elena P6rez-Mifiana, Peter Ross, John Hallam A Genetic Algorithm for Planning Coal Purchase of a Real Electric Power Plant Masahiro Inuiguchi, Tadahiro Miyake, Masatoshi Sakawa, Isao Shiromaru "'CAM-BRAIN" ATR~ Billion Neuron Artificial Brain Project -A Three Year Progress Report- Hugo de Garis