PRICAI 2002: Trends in Artificial Intelligence: 7th Pacific Rim International Conference on Artificial Intelligence, Tokyo, Japan, August 18-22, 2002. ... (Lecture Notes in Computer Science, 2417)

The Paci?c Rim International Conferences on Arti?cial Intelligence (PRICAI) are biennial international events. The PRICAI series aims at stimulating - search by promoting exchange and cross-fertilization of ideas among di?erent branches of Arti?cial Intelligence. It also provides a common forum for rese- chers and practitioners in various ?elds of AI to exchange new ideas and share their experience. This volume contains the proceedings of the 7th Paci?c Rim Internat- nal Conference on Arti?cial Intelligence (PRICAI 2002) heldin Tokyo, Japan. PRICAI 2002 received161 submissions from 25 countries. From these, 57 papers (35%) were acceptedfor presentation andare includedin this volume. Over 30 papers were acceptedfor poster presentations, anda 1-page abstract for each poster is includedin this volume. All submittedpapers were refereedby two or more reviewers selectedby the Program Committee members. The reviewers’ comments were carefully examinedanddiscussedby the members of the P- gram Committee to ensure fairness andconsistency in the selection process. The papers in this volume give an indication of new trends in Arti?cial Intelligence. The topics covered include AI Foundations, Knowledge Repres- tation andReasoning, Constraint Satisfaction, Machine Learning, Knowledge Acquisition andManagement, Agents andTheir Applications, Neural Networks, Document Analysis andCategorization, Internet/Web Intelligence, Multimedia andEmotion, Intelligent Robots, andMachine Perception. PRICAI 2002: Trends in Artificial Intelligence Preface Organization Table of Contents Commercial Applications of Machine Learning for Personalized Wireless Portals The Need for Adaptive Personalization Machine Learning of User Models Applications of Adaptive Personalization Conclusion References Learning, Collecting, and Using Ontological Knowledge for NLP Hidden Variables in Knowledge Representation Extended Summary Succinctness and Coherence Philosophical Precedents Model Theory Nondeterminism What Is a Representation? Intellimedia Systems: Research and Applications at the Intersection of Multimedia and Artificial Intelligence Introduction Presenting Information through Life-Like Characters Animated Presenters Employing Multiple Agents in Presentation Tasks Intellimedia for Mobile Users Customizing Graphics for Tiny Displays of Mobile Devices Media and Content Management in an Intelligent Driver Support System References Integration of Multimedia and Art for New Human-Computer Communications Introduction Trends of Communication Media Multimedia and Future Communication Media Nonverbal Aspect of Communications Approach Aiming at the Integration of Art and Technology MIDAS: Mic Intreractive Dance System Overview Overview of MIDAS Method of Extracting Emotion from Human Body Motion Future Work Costune System Design and Prototype Discussion Overview Conclusion References R-UNSEARCHMO: A Refinement on UNSEARCHMO Introduction SATCHMO and UNSEARCHMO R-UNSEARCHMO: A Refinement on UNSEARCHMO Implementation Experimental Results Conclusion References Deontic Relevant Logic: A Strong Relevant Logic Approach to Removing Paradoxes from Deontic Logic Introduction Paradoxes in Deontic Logic Strong Relevant Logic Deontic Relevant Logic Comparison with Related Works Conclusion References Solving the Ramification Problem: Causal Propagation in an Argumentation-Theoretic Approach Introduction and Motivation Preliminaries and Background Syntax and Semantics Background Technical Framework Connection to Thielscher's (1997) Causal Representation Running Examples Conclusion References Representing Actions over Dynamic Domains Introduction Syntax of ${cal D}$ Semantics of ${cal D}$ $d$-Structures Transition Function, Models, and Entailment Semantic Properties of Domain Descriptions Further Discussions References Consistency of Action Descriptions Introduction EPDL Preliminaries Action Descriptions and Their Normal Forms Normal Action Descriptions Consistency of Action Descriptions Uniform Consistency of Action Descriptions $Sigma $-Consistency of Formulas Regional Consistency of Action Descriptions Conclusion References Solving Factored MDPs with Large Action Space Using Algebraic Decision Diagrams Introduction FMDPs with Factored Action Spaces (FA-FMDPs) Model Minimization Algorithm for FA-FMDPs Using Algebraic Decision Diagrams (ADDs) Experiments Discussion and Related Work References Dynamic Fuzziness Introduction Dynamic Fuzzy Sets Dynamic Fuzzy Calculus Computational Example Conclusions References Distributed Reinforcement of Arc-Consistency Introduction DRAC Architecture DRAC Dynamic Communication Protocol Data Structures Basic Primitives Global Dynamic Agent Behaviors Example Discussions Experimentation Conclusion References Parallel Execution of Stochastic Search Procedures on Reduced SAT Instances Introduction The Basic Approach The Parallel Approach Related Works Conclusion References Two Transformations of Clauses into Constraints and Their Properties for Cost-Based Hypothetical Reasoning Introduction How to Make Constraints? Preprocessing Transformation into Linear Constraints Transformation into Nonlinear Constraints Discussion of Two Transformations Difference of Two Transformations Performance Transition of Transformation L Performance of Transformation {bf NL} Conclusion References Hidden Markov Modeling for Multi-agent Systems Introduction HMM for Agents with Dynamic Environment Agent Model with Environment Environment as Output Symmetrically Coupled HMM Symmetricity Assumption Formalization and Learning Procedure Discussion: The Number of Parameters in the Model Related Works Concluding Remarks References Modelling PRS-Like Agents' Mental States Introduction Computation Tree Logic BDI Logic Agent Dynamic Logic Conclusion References Genetic Algorithm and Social Simulation Introduction Repeated-Evaluation Genetic Algorithm Artificial Society Chromosome Representation Experimentation and Results Conclusions References Adaptive Directed Acyclic Graphs for Multiclass Classification Introduction Decision Directed Acyclic Graphs The DDAG Architecture & the DAGSVM Algorithm Issues on DDAG Adaptive Directed Acyclic Graphs The ADAG Architecture Implementation Analyses of DDAG & ADAG Experiments The Thai Vowel Data Set The UCI Letter Data Set Conclusion Appendix References Network Optimization through Learning and Pruning in Neuromanifold Introduction Hierarchical Structure of Neuromanifold Network Optimization in Neuromanifold Geometrical Descriptions of Network Optimization Natural Gradient Learning Method Natural Pruning Experimental Results Conclusions and Discussions References DIC: A Novel Discrete Incremental Clustering Technique for the Derivation of Fuzzy Membership Functions Introduction The GenSoFNN Network The Discrete Incremental Clustering (DIC) Technique The Fuzzy Set Support Parameter SLOPE Plasticity Parameter beta Tendency Parameter TD Thresholds (IT and OT) Experimental Results Conclusions References Application of Episodic Q-Learning to a Multi-agent Cooperative Task Introduction $N$-Hunter Game Q-Learning and PS Learning Episodic Q-Learning Restricting the History Length Discussions --- The Algorithm to Read Other's Mind References LC-Learning: Phased Method for Average Reward Reinforcement Learning - Analysis of Optimal Criteria Introduction Markov Decision Process and Policy Analysis of Policy Optimality Necessary and Sufficient Condition of Gain-Optimality Disambiguation for Bias-Optimality Foundation of LC-Learning Detect all Cycles Select Optimal Cycle Evaluate Cycle-Bias-Value Evaluate Bias-Value Conclusion Related Work References LC-Learning: Phased Method for Average Reward Reinforcement Learning - Preliminary Results - Introduction Bus Scheduling Task LC-Learning Framework Pruning Preliminary Results Dynamic Value Domain Comparative Experiments with Prioritized Sweeping Conclusion Future Work References Extension of the RDR Method That Can Adapt to Environmental Changes and Acquire Knowledge from Both Experts and Data Introduction Ripple Down Rules Revisited The Minimum Description Length Principle The DL of a Binary Tree The DL for Class Labels of Misclassified Cases Knowledge Acquisition Based on MDLP Knowledge Acquisition from Data Alone Knowledge Acquisition from Both Data and Human Experts The Knowledge Deletion for the Environmental Changes The Criterion to Estimate the Value of Knowledge Deletion of Node from a Binary Tree Experiment Results and Discussions Conclusion References Case Generation Method for Constructing an RDR Knowledge Base Introduction Ripple Down Rules Concept for Case Generation Six Algorithms for Generating Cases Knowledge Aquisition in RDR Six Algorithms for Generating Cases Experiment and Discussion Experiment Results and Discussions Conclusion References Association Rules Using Rough Set and Association Rule Methods Introduction Mining Multiple Level Association Rule (MML-AR) Model Pre-processing Module Data Transformation Data Cleaning Processing Module Rule Generation Rule Evaluation Experimental Results Conclusion References Change-Point Estimation Using New Minimum Message Length Approximations Introduction Binomial Problem The Minimum Message Length Principle Strict Minimum Message Length (SMML) Fairly Strict Minimum Message Length (FSMML) MML68 Change-Point Approximation A New Approximation to FSMML: MMLD Empirical Comparison and Discussion Conclusion References Knowledge Discovery from Structured Data by Beam-Wise Graph-Based Induction Introduction Graph-Based Induction GBI Revisited Extracting Discriminative Patterns Beam-Wise Graph-Based Induction Algorithm of B-GBI Canonical Labeling Experimental Evaluation of B-GBI Conclusion References BackPOLE: Back Propagation Based on Objective Learning Errors Introduction The GenSoFNN Network NGD-BP Based Parameter Learning of GenSoFNN Shortcomings of NGD-BP in GenSoFNN The Proposed BackPOLE Algorithm Experimental Results Conclusions References A Method on Improvement of the Online Mode Error Backpropagation Algorithm for Pattern Recognition Introduction Analysis of the Offline and Online EBP Learning Using the Repulsive Force Concept Possibility of Improving the Online EBP Learning Speed Proposals Experimentations Conclusion References Optimizing a Multiple Classifier System Introduction Self-Generating Neural Networks Optimization of a Multiple Classifier System Experimental Results Conclusions References Generalization of Iterative Learning Control for Multiple Desired Trajectories in Robotic Systems Introduction Problem Formulation Generalization of Experience for Multiple Desired Trajectories Selection of Initial Control Input NN as a Universal Function Approximator Modelling of Experience Convergence Property Numerical Illustration –An Example of Two Link Robot Conclusions References Learning Topological Maps from Sequential Observation and Action Data under Partially Observable Environment Introduction Problem Definition Observation Data and Map Model Model Likelihood and Entropies as Criteria Globally Distinctive States Proposed Method Overview Discretization of Observation Space by Distance-Based Clustering State Splitting Based on Transition Probability Distribution Simulation Assumed Environment and Robot Clustering of Observation Vectors State Splitting Conclusion References A Saliency Map Model for Active Perception Using Color Information and Local Competitive Mechanism Introduction The Saliency Map Model Feature Extraction Color Processing in Human Visual System The Feature Maps Feature Combination The Need for the Saliency Map Combination Process Experimental Results Discussion and Conclusions References Generation of Optimal Biped Walking for Humanoid Robot by Co-evolving Morphology and Controller Introduction Materials and Methods Morphology Controller Method Simulation Evolutionary Computation Results and Discussions Multi Optimal Problem Results and Disccusions Conclusions References Multi-agent Coordination in Planning Introduction A Framework for Multi-agent Planning Plan Coordination Plan Coordination by Partitioning Local Plans Partitioning Using Only Local Dependency Information A Refined Partitioning Strategy Conclusions References A Multi-agent Based Approach to the Inventory Routing Problem Introduction The Inventory Routing Problem Our Multi-agent Approach to IRP Agent Coordination and Control Experiment Results Conclusion References An Agent-Based Hybrid Intelligent System for Financial Investment Planning Introduction Description of the Investment Planning Problem The Agent Architecture for Investment Planning Implementation of the System Experimental Results Comparison with Related Work and Evaluations Concluding Remarks References Socially Intelligent Aerial Combat Simulator Introduction Team Roles Role Hierarchy Role Transitions and Exchanges Experimental Setup and Analysis Experiment Parameters Result Analysis Related Work Conclusion and Future Work References Construction of Large-Scale Bayesian Networks by Local to Global Search Introduction The Markov Blanket Structure The ``Local to Global Search" Algorithm Experimental Evaluation Conclusion References Using Bayesian Networks with Hidden Nodes to Recognise Neural Cell Morphology Introduction Hidden Node Methodology Formalisation of Hidden Node Methodology Applications to the Morphological Classification of Neural Cells Naïve Bayesian Networks Training Hidden Node(s) Summary and Discussion References Recognizing 100 Speakers Using Homologous Naive Bayes Introduction Naive Bayesian Classifiers Classifying a Homologous Set Application in Speaker Recognition Gaussian Mixture Model Experiments Conclusions and Future Work References An Approach to Microscopic Clustering of Terms and Documents Introduction Mathematical Development Definition of Micro-Clusters Entropy-Based Criteria for Cluster Evaluation Strategy for Cluster Generation Cluster Initiation Local Improvement and Testing Experimental Results Conditions for Experiments Using NTCIR-J1 Clustering Results Future Issues References Effective Methods for Improving Naive Bayes Text Classifiers Introduction Multinomial Naive Bayes for Text Classifier Our Approach Document Model Based Parameter Estimation Document Length Normalization MI-Weighted Naive Bayes Classifier Experimental Results Conclusion and Future Work References Efficiently Clustering Documents with Committees Introduction Related Work Representation Algorithm Phase I: Find Top-Similar Elements Phase II: Find Committees Phase III: Assign elements to Clusters Evaluation Methodology Experimental Results Test Data Cluster Evaluation K-Means and Buckshot Conclusion References Topic Extraction from Text Documents Using Multiple-Cause Networks Introduction The Multiple-Cause Model Multiple-Cause Networks The Helmholtz Machines Applying to Text Documents Experiments Topic Words Extraction Document Clustering Conclusions References A Comparative Study on Statistical Machine Learning Algorithms and Thresholding Strategies for Automatic Text Categorization 1 Introduction 2 Reviews on Common Techniques 2.1 Representation and Feature Selection 2.2 Statistical Learning Algorithms: Rocchio and k-NN 2.3 Common Thresholding Strategies 3 Keyword Association Network (KAN) 3.1 Need for New Learning Algorithm 3.2 New Representation and Learning Algorithm: KAN Definition 1 Definition 2 Definition 3 4 New Thresholding Strategy: RinSCut 5 Experiments 5.1 Data Sets 5.2 Experimental Setup 5.3 Results 6 Conclusions References PATI: An Approach for Identifying and Resolving Ambiguities Introduction Definition of PATI Grammar Tuning Rule Preference Function and Overall Scoring Scheme Experiments Conclusion References Extracting User Profiles from E-mails Using the Set-Oriented Classifier Introduction Related Works SCS (Set-Oriented Classifier System) The Algorithm Fitness Function Experiments and Results Conclusion References Wrapper Generation by Using XML-Based Domain Knowledge for Intelligent Information Extraction Introduction Domain Knowledge Wrapper Generation Implementation and Evaluation Conclusion References Modified PrefixSpan Method for Motif Discovery in Sequence Databases Introduction Problem Definition and PrefixSpan Method New Method Modified PrefixSpan Method Extracting Frequent Patterns Experimental Results Performance Evaluation Question and Answering Interface Conclusion References A Multi-agent Bioinformatics Integration System with Adjustable Autonomy Introduction Architecture Overview Decision Modes Prototype System Initialization Protocols Ontologies Decision Criticality Explanation Conclusion References Using Case-Based Reasoning Approach in Planning Instructional Activities Introduction Representation and Organization of Cases Case Representation and Indexing Case Organization Retrieval and Adaptation Case Retrieval Case Evaluation, Repair, and Learning Case Learning Summary and Conclusion References Feature Construction for Student Group Forming Based on Their Browsing Behaviors in an E-learning System Introduction Curriculum Knowledge in Our Domain Simulator Feature Construction A Formal Description of Feature Construction An Arithmetic-Composition-Based Feature Construction Experimentation Verifications A Weighted-Euclidean-Distance-Based Data Clustering Conclusions and Future Work References Web-Based Tutoring System for Computer Security Introduction System Design The Learning Environment ITS Structure The Virtual Simulation Task Generation Procedural Event Evaluation Implementation Sample Tracing Scenario Conclusion References A Simple Illumination Normalization Algorithm for Face Recognition Introduction Multiple Regression Model Illumination C ompensation Based on the Multiple R egression Model (ICR) Background ICR Algorithm ICR Results Experimental Results Geometric Normalization Test Database Results Conclusions References A Self-Adaptive Architecture and Its Application to Robust Face Identification Introduction The Face Recognition Problem An Overview of the GRAVA Architecture Reflective Interpreter for Self-Adaptation Protocol for Interpreters Protocol for Agents Conclusions References Realizing Audio-Visually Triggered ELIZA-Like Non-verbal Behaviors Introduction Related Work Humanoid Hardware Perceptual Systems in Real-Time Multiple-Talker Tracking Active Audition Module Face Recognition and Identification Module Stream Formation and Association Attention System Socially-Oriented Attention Control Task-Oriented Attention Control Experiments and Observation Task-Oriented Interaction: {it SIG/} as a Receptionist Robot Socially-Orineted Interaction: {it SIG/} as a Companion Robot Observation: {it SIG} as a Non-verbal Eliza Conclusions References Audio-to-Visual Conversion Using Hidden Markov Models Introduction Visemes Viseme Recognition Experimental Results Conclusions References Scripting the Bodies and Minds of Life-Like Characters Introduction The SCREAM System Emotion Generation Emotion Regulation Emotion Expression Affect Processing MPML: A Markup Language for Character Control Illustrative Example Conclusion References An Affective Decision Making Agent Architecture Using Emotion Appraisals Introduction The EMAI Architecture EMAI Ontology Goal Hierarchy Event and Plan Definitions Affective Space Event Appraisal Affective Decision Making Results of Experimentation with an EMAI Agent Conclusions and Further Work References Logic Programming for Agents References Evolutionary Multi-agents System Agent-Based Cooperative Distributed Tutoring Systems A Computational Model of Reasoning as Socially-Constructed Process References Managing Information Complexity of Supply Chains via Agent-Based Genetic Programming References Semantic Integration of E-business Models Based on Multi-layered Repository Overview References Randomization and Uncertain Inference References Checkers Strategy Evolution with Speciated Neural Networks Generation and Optimization of Fuzzy Neural Network Structure FuzzyDrive: A Fuzzy Rule-Based Auto Pilot System Implementing NRDR Using OO Database Management System (OODBMS) Implementing NRDR Using F2 References Using Ripple Down Rules for Actions and Planning A Generative Dependency Grammar References Indonesian Morphological Parser with Minimum Connectivity Cost to Solve Ambiguities References Target Word Selection Using WordNet and Data-Driven Models in Machine Translation A Study on Using Natural Language as a Computer Communication Protocol Automatic Indexing Based on Term Activities References Answer Extraction by Flexible Matching, Filtering, and Interpretation A Statistical Identification and Verification Method for Biometrics Proposal of a Multimodal Dialogue Description Language Image Classification by Web Images References Real-Time Face Detection and Tracking Using PCA and NN The Proposed Method A Wrapper-Based Approach to Robot Learning Concepts from Images References An Effective HMM-Based Intrusion Detection System with Privilege Change Event Modeling A Method on Improving of Enrolling Speed for the MLP-Based Speaker Verification System through Reducing Learning Data References Syntactic Representations of Semantic Merging Operations References Author Index

This book constitutes the refereed proceedings of the 7th Pacific Rim International Conference on Artificial Intelligence, PRICAI 2002, held in Tokyo, Japan in August 2002.
The 57 revised full papers presented together with 5 invited contributions and 26 posters were carefully reviewed and selected from 161 submissions. The papers are organized in topical sections on logic and AI foundations, representation and reasoning of actions, constraint satisfaction, foundations of agents, foundations of learning, reinforcement learning, knowledge acquisition and management, data mining and knowledge discovery, neural network learning, learning for robots, multi-agent applications, document analysis, Web intelligence, bioinformatics, intelligent learning environments, face recognition, and multimedia and emotion.