Learning Horn definitions: Theory and an application to planning

A Horn definition is a set of Horn clauses with the same predicate in all head literals. In this paper, we consider learning non-recursive, first-order Horn definitions from entailment. We show that this class is exactly learnable from equivalence and membership queries. It follows then that this class is PAC learnable using examples and membership queries. Finally, we apply our results to learning control knowledge for efficient planning in the form of goal-decomposition rules. Chandra Reddy, Ph.D.: He is currently a doctoral student in the Department of Computer Science at Oregon State University. He is completing his Ph.D. on June 30, 1998. His dissertation is entitled “Learning Hierarchical Decomposition Rules for Planning: An Inductive Logic Programming Approach.” Earlier, he had an M. Tech in Artificial Intelligence and Robotics from University of Hyderabad, India, and an M.Sc.(tech) in Computer Science from Birla Institute of Technology and Science, India. His current research interests broadly fall under machine learning and planning/scheduling—more specifically, inductive logic programming, speedup learning, data mining, and hierarchical planning and optimization. Prasad Tadepalli, Ph.D.: He has an M.Tech in Computer Science from Indian Institute of Technology, Madras, India and a Ph.D. from Rutgers University, New Brunswick, USA. He joined Oregon State University, Corvallis, as an assistant professor in 1989. He is now an associate professor in the Department of Computer Science of Oregon State University. His main area of research is machine learning, including reinforcement learning, inductive logic programming, and computational learning theory, with applications to classification, planning, scheduling, manufacturing, and information retrieval.     

Cooperation of categorical and behavioral learning in a practical solution to the abstraction problem

Real robots should be able to adapt autonomously to various environments in order to go on executing their tasks without breaking down. They achieve this by learning how to abstract only useful information from a huge amount of information in the environment while executing their tasks. This paper proposes a new architecture which performs categorical learning and behavioral learning in parallel with task execution. We call the architectureSituation Transition Network System (STNS). In categorical learning, it makes a flexible state representation and modifies it according to the results of behaviors. Behavioral learning is reinforcement learning on the state representation. Simulation results have shown that this architecture is able to learn efficiently and adapt to unexpected changes of the environment autonomously. Atsushi Ueno, Ph.D.: He is a research associate in the Artificial Intelligence Laboratory at the Graduate School of Information Science at the Nara Institute of Science and Technology (NAIST). He received the B.E., the M.E., and the Ph.D. degrees in aeronautics and astronautics from the University of Tokyo in 1991, 1993, and 1997 respectively. His research interest is robot learning and autonomous systems. He is a member of Japan Association for Artificial Intelligence (JSAI). Hideaki Takeda, Ph.D.: He is an associate professor in the Artificial Intelligence Laboratory at the Graduate School of Information Science at the Nara Institute of Science and Technology (NAIST). He received his Ph.D. in precision machinery engineering from the University of Tokyo in 1991. He has conducted research on a theory of intelligent computer-aided design systems, in particular experimental study and logical formalization of engineering design. He is also interested in multiagent architectures and ontologies for knowledge base systems.     

Using discriminant analysis for multi-class classification: an experimental investigation

Many supervised machine learning tasks can be cast as multi-class classification problems. Support vector machines (SVMs) excel at binary classification problems, but the elegant theory behind large-margin hyperplane cannot be easily extended to their multi-class counterparts. On the other hand, it was shown that the decision hyperplanes for binary classification obtained by SVMs are equivalent to the solutions obtained by Fisher's linear discriminant on the set of support vectors. Discriminant analysis approaches are well known to learn discriminative feature transformations in the statistical pattern recognition literature and can be easily extend to multi-class cases. The use of discriminant analysis, however, has not been fully experimented in the data mining literature. In this paper, we explore the use of discriminant analysis for multi-class classification problems. We evaluate the performance of discriminant analysis on a large collection of benchmark datasets and investigate its usage in text categorization. Our experiments suggest that discriminant analysis provides a fast, efficient yet accurate alternative for general multi-class classification problems. Tao Li is currently an assistant professor in the School of Computer Science at Florida International University. He received his Ph.D. degree in Computer Science from University of Rochester in 2004. His primary research interests are: data mining, machine learning, bioinformatics, and music information retrieval. Shenghuo Zhu is currently a researcher in NEC Laboratories America, Inc. He received his B.E. from Zhejiang University in 1994, B.E. from Tsinghua University in 1997, and Ph.D degree in Computer Science from University of Rochester in 2003. His primary research interests include information retrieval, machine learning, and data mining. Mitsunori Ogihara received a Ph.D. in Information Sciences at Tokyo Institute of Technology in 1993. He is currently Professor and Chair of the Department of Computer Science at the University of Rochester. His primary research interests are data mining, computational complexity, and molecular computation.     

Evolvable hardware: A robot navigation system testbed

Recently there has been great interest in the design and study of evolvable systems based on Artificial Life principles in order to monitor and control the behavior of physically embedded systems such as mobile robots, plants and intelligent home devices. At the same time new integrated circuits calledsoftware-reconfigurable devices have been introduced which are able to adapt their hardware almost continuously to changes in the input data or processing. When the configuration phase and the execution phase are concurrent, the software-reconfigurable device is calledevolvable hardware (EHW). This paper examines an evolutionary navigation system for a mobile robot using a Boolean function approach implemented on gate-level evolvable hardware (EHW). The task of the mobile robot is to reach a goal represented by a colored ball while avoiding obstacles during its motion. We show that the Boolean function approach using dedicated evolution rules is sufficient to build the desired behavior and its hardware implementation using EHW allows to decrease the learning time for on-line training. We demonstrate the effectiveness of the generalization ability of the Boolean function approach using EHW due to its representation and evolution mechanism. The results show that the evolvable hardware configuration learned off-line in a simple environment creates a robust robot behavior which is able to perform the desired behaviors in more complex environments and which is insensitive to the gap between the real and simulated world. Didier Keymeulen, Ph.D.: He currently works as a senior research engineer at the Computer Science Division of Electrotechnical Laboratory, AIST, MITI, Japan. His research interests are in the design of adaptive physically embedded systems using biologically inspired complex dynamical systems. He studied electrical and computer science engineering at the Universite Libre de Bruxelles in 1987. He obtained his M. Sc. and PH. D. in Computer Science from the Artificial Intelligence Laboratory of the Vrije Universiteit Brussel, directed by Dr. Luc Steels, respectively in 1991 and 1994. He was the Belgium laureate of the Japanese JSPS Postdoctoral Fellowship for Foreign Researchers in 1995. Masaya Iwata, Ph.D.: He currently works as a researcher at the Computer Science Division of Electrotechnical Laboratory, AIST, MITI, Japan. His research interests are in developing adaptive hardware devices using genetic algorithms, and in their applications to pattern recognition and image compression. He received his B. E. in 1988, his M. E. in 1990, and his Ph. D. in 1993 in applied physics from the Osaka University. He was a postdoctoral fellow in optical computing at ONERA-CERT, Toulouse, France in 1993. Kenji Konaka: He is currently working as a software research engineer at the Humanoid Interaction Laboratory of the Intelligent Systems Division of Electrotechnical Laboratory, AIST, MITI, Japan. His current research interest is on real-time vision-based mobile robots working in cooperative mode. He has developped a highly interactive distributed real-time software and hardware platform for controlling a group of robots. Yasuo Kuniyoshi, Ph.D.: He is currently a senior research scientist and head of the Humanoid Interaction Laboratory at the Intelligent Systems Division of Electrotechnical Laboratory, AIST, MITI, Japan. His current research interest is on emergence of stable structures out of complex sensory-motor interactions by a humanoid robot. He received IJCAI93 Outstanding Paper A ward and several other awards in the field of intelligent robotics. He received the B. Eng. in applied physics in 1985, M. Eng. and Ph. D. in information engineering in 1988 and 1991 respectively, all from the University of Tokyo. Tetsuya Higuchi, Ph.D.: He heads the Evolvable Systems Laboratory in Electrotechnical Laboratory, AIST, MITI, Japan. He received B. E., M. E., Ph. D. degrees all in electrical engineering from Keio University in 1978, 1980, and 1984, respectively. His current interests include envolvable hardware systems, parallel processing architecture in artificial intelligence, and adaptive systems. He is also in charge of the adaptive devices group in the MITI national project, Real World Computing Project.     

Belief revision via Lamarckian evolution

We present a system for performing belief revision in a multi-agent environment. The system is called GBR (Genetic Belief Revisor) and it is based on a genetic algorithm. In this setting, different individuals are exposed to different experiences. This may happen because the world surrounding an agent changes over time or because we allow agents exploring different parts of the world. The algorithm permits the exchange of chromosomes from different agents and combines two different evolution strategies, one based on Darwin’s and the other on Lamarck’s evolutionary theory. The algorithm therefore includes also a Lamarckian operator that changes the memes of an agent in order to improve their fitness. The operator is implemented by means of a belief revision procedure that, by tracing logical derivations, identifies the memes leading to contradiction. Moreover, the algorithm comprises a special crossover mechanism for memes in which a meme can be acquired from another agent only if the other agent has “accessed” the meme, i.e. if an application of the Lamarckian operator has read or modified the meme. Experiments have been performed on the η-queen problem and on a problem of digital circuit diagnosis. In the case of the η-queen problem, the addition of the Lamarckian operator in the single agent case improves the fitness of the best solution. In both cases the experiments show that the distribution of constraints, even if it may lead to a reduction of the fitness of the best solution, does not produce a significant reduction. Evelina Lamma, Ph.D.: She is Full Professor at the University of Ferrara. She got her degree in Electrical Engineering at the University of Bologna in 1985, and her Ph.D. in Computer Science in 1990. Her research activity centers on extensions of logic programming languages and artificial intelligence. She was coorganizers of the 3rd International Workshop on Extensions of Logic Programming ELP92, held in Bologna in February 1992, and of the 6th Italian Congress on Artificial Intelligence, held in Bologna in September 1999. Currently, she teaches Artificial Intelligence and Fondations of Computer Science. Fabrizio Riguzzi, Ph.D.: He is Assistant Professor at the Department of Engineering of the University of Ferrara, Italy. He received his Laurea from the University of Bologna in 1995 and his Ph.D. from the University of Bologna in 1999. He joined the Department of Engineering of the University of Ferrara in 1999. He has been a visiting researcher at the University of Cyprus and at the New University of Lisbon. His research interests include: data mining (and in particular methods for learning from multirelational data), machine learning, belief revision, genetic algorithms and software engineering. Luís Moniz Pereira, Ph.D.: He is Full Professor of Computer Science at Departamento de Informática, Universidade Nova de Lisboa, Portugal. He received his Ph.D. in Artificial Intelligence from Brunel University in 1974. He is the director of the Artificial Intelligence Centre (CENTRIA) at Universidade Nova de Lisboa. He has been elected Fellow of the European Coordinating Committee for Artificial Intelligence in 2001. He has been a visiting Professor at the U. California at Riverside, USA, the State U. NY at Stony Brook, USA and the U. Bologna, Italy. His research interests include: knowledge representation, reasoning, learning, rational agents and logic programming.     

A new approach to fuzzy modeling and control for nonlinear dynamic systems: Neuro-fuzzy dynamic characteristic modeling and adaptive control mechanism

The study on nonlinear control system has received great interest from the international research field of automatic engineering. There are currently some alternative and complementary methods used to predict the behavior of nonlinear systems and design nonlinear control systems. Among them, characteristic modeling (CM) and fuzzy dynamic modeling are two effective methods. However, there are also some deficiencies in dealing with complex nonlinear system. In order to overcome the deficiencies, a novel intelligent modeling method is proposed by combining fuzzy dynamic modeling and characteristic modeling methods. Meanwhile, the proposed method also introduces the low-level learning power of neural network into the fuzzy logic system to implement parameters identification. This novel method is called neuro-fuzzy dynamic characteristic modeling (NFDCM). The neuro-fuzzy dynamic characteristic model based overall fuzzy control law is also discussed. Meanwhile the local adaptive controller is designed through the golden section adaptive control law and feedforward control law. In addition, the stability condition for the proposed closed-loop control system is briefly analyzed. The proposed approach has been shown to be effective via an example. Recommended by Editor Young-Hoon Joo. This work was jointly supported by National Natural Science Foundation of China under Grant 60604010, 90716021, and 90405017 and Foundation of National Laboratory of Space Intelligent Control of China under Grant SIC07010202. Xiong Luo received the Ph.D. degree from Central South University, Changsha, China, in 2004. From 2005 to 2006, he was a Postdoctoral Fellow in the Department of Computer Science and Technology at Tsinghua University. He currently works as an Associate Professor in the Department of Computer Science and Technology, University of Science and Technology Beijing. His research interests include intelligent control for spacecraft, intelligent optimization algorithms, and intelligent robot system. Zengqi Sun received the bachelor degree from Tsinghua University, Beijing, China, in 1966, and the Ph.D. degree from Chalmers University of the Technology, Gothenburg, Sweden, in 1981. He currently works as a Professor in the Department of Computer Science and Technology, Tsinghua University. His research interests include intelligent control of robotics, fuzzy neural networks, and intelligent flight control. Fuchun Sun received the Ph.D. degree from Tsinghua University, Beijing, China, in 1998. From 1998 to 2000, he was a Postdoctoral Fellow in the Department of Automation at Tsinghua University, where he is currently a Professor in the Department of Computer Science and Technology. His research interests include neural-fuzzy systems, variable structure control, networked control systems, and robotics.     

Advantages of decision lists and implicit negatives in Inductive Logic Programming

This paper demonstrates the capabilities offoidl, an inductive logic programming (ILP) system whose distinguishing characteristics are the ability to produce first-order decision lists, the use of an output completeness assumption as a substitute for negative examples, and the use originally motivated by the problem of learning to generate the past tense of English verbs; however, this paper demonstrates its superior performance on two different sets of benchmark ILP problems. Tests on the finite element mesh design problem show thatfoidl’s decision lists enable it to produce generally more accurate results than a range of methods previously applied to this problem. Tests with a selection of list-processing problems from Bratko’s introductory Prolog text demonstrate that the combination of implicit negatives and intensionality allowfoidl to learn correct programs from far fewer examples thanfoil. This research was supported by a fellowship from AT&T awarded to the first author and by the National Science Foundation under grant IRI-9310819. Mary Elaine Califf: She is currently pursuing her doctorate in Computer Science at the University of Texas at Austin where she is supported by a fellowship from AT&T. Her research interests include natural language understanding, particularly using machine learning methods to build practical natural language understanding systems such as information extraction systems, and inductive logic programming. Raymond Joseph Mooney: He is an Associate Professor of Computer Sciences at the University of Texas at Austin. He recerived his Ph.D. in Computer Science from the University of Illinois at Urbana-Champaign in 1988. His current research interests include applying machine to natural language understanding, inductive logic programming, knowledge-base and theory refinement, learning for planning, and learning for recommender systems. He serves on the editorial boards of the journalNew Generation Computing, theMachine Learning journal, theJournal of Artificial Intelligence Research, and the journalApplied Intelligence.     

Constructive negation and constraint logic programming with sets

The aim of this paper is to extend theConstructive Negation technique to the case ofCLP(SεT), a Constraint Logic Programming (CLP) language based on hereditarily (and hybrid) finite sets. The challenging aspects of the problem originate from the fact that the structure on whichCLP(SεT) is based is notadmissible closed, and this does not allow to reuse the results presented in the literature concerning the relationships betweenCLP and constructive negation. We propose a new constraint satisfaction algorithm, capable of correctly handling constructive negation for large classes ofCLP(SεT) programs; we also provide a syntactic characterization of such classes of programs. The resulting algorithm provides a novel constraint simplification procedure to handle constructive negation, suitable to theories where unification admits multiple most general unifiers. We also show, using a general result, that it is impossible to construct an interpreter forCLP(SεT) with constructive negation which is guaranteed to work for any arbitrary program; we identify classes of programs for which the implementation of the constructive negation technique is feasible. Agostino Dovier, Ph.D.: He is a researcher in the Department of Science and Technology at the University of Verona, Italy. He obtained his master degree in Computer Science from the University of Udine, Italy, in 1991 and his Ph.D. in Computer Science from the University of Pisa, Italy, in 1996. His research interests are in Programming Languages and Constraints over complex domains, such as Sets and Multisets. He has published over 20 research papers in International Journals and Conferences. He is teaching a course entitled “Special Languages and Techniques for Programming” at the University of Verona. Enrico Pontelli, Ph.D.: He is an Assistant Professor in the Department of Computer Science at the New Mexico State University. He obtained his Laurea degree from the University of Udine (Italy) in 1991, his Master degree from the University of Houston in 1992, and his Ph.D. degree from New Mexico State University in 1997. His research interests are in Programming Languages, Parallel Processing, and Constraint Programming. He has published over 50 papers and served on the program committees of different conferences. He is presently the Associate Director of the Laboratory for Logic, Databases, and Advanced Programming. Gianfranco Rossi, Ph.D.: He received his degree in Computer Science from the University of Pisa in 1979. From 1981 to 1983 he was employed at Intecs Co. System House in Pisa. From November 1983 to February 1989 he was a researcher at the Dipartimento di Informatica of the University of Turin. Since March 1989 he is an Associate Professor of Computer Science, currently with the University of Parma. He is the author of several papers dealing mainly with programming languages, in particular logic programming languages and Prolog, and extended unification algorithms. His current research interests are (logic) programming languages with sets and set unification algorithms.     

Relative positioning for team robot navigation

The research presented in this paper approaches the issue of robot team navigation using relative positioning. With this approach each robot is equipped with sensors that allow it to independently estimate the relative direction of an assigned leader. Acoustic sensor systems are used and were seen to work very effectively in environments where datum relative positioning systems (such as GPS or acoustic transponders) are typically ineffective. While acoustic sensors provide distinct advantages, the variability of the acoustic environment presents significant control challenges. To address this challenge, directional control of the robot was accomplished with a feed forward neural network trained using a genetic algorithm, and a new approach to training using recent memories was successfully implemented. The design of this controller is presented and its performance is compared with more traditional classic logic and behavior controllers. Patrick McDowell received his bachelor's degree in Computer Science in 1984 from the University of Idaho. He spent the next 15 years working as a computer scientist for a small defense contractor where he specialized in real time data acquisition, application development, and image processing. In 1999 he received his master's degree in computer science from the University of Southern Mississippi. In 2000 he began work at the Naval Research Lab where he has focused on application of machine learning techniques to autonomous underwater navigation. In 2005 he received his Ph.D. in Computer Science from Louisiana State University. His research interests include legged robotics, machine learning, and artificial intelligence. In Fall of 2006 he joined Southeastern Louisiana University as an assistant professor of Computer Science. Brian S. Bourgeois received his Ph.D. in Electrical Engineering from Tulane University located in New Orleans, LA in 1991. Since then he has worked at the Stennis Space Center, MS detachment of the Naval Research Laboratory. He has worked on research projects spanning an array of technologies including airborne survey sytems, acoustic backscattering, bathymetry and imaging sonar systems, the ORCA unmanned underwater vehicle and the development of an autonomous survey system for hydrographic survey ships. He is presently the head of the Position, Navigation and Timing team at NRL with research interests including underwater positioning and communications and autonomous navigation. Ms. McDowell received her M.S. in Applied Physics in 2002 from the University or New Orleans. She is presently a candidate for a Ph. D. in Engineering and Applied Science. She joined the Naval Research Laboratory in 1991 as a research engineer and has spent most of that time working in experimental and theoretical acoustic modeling. Ms. McDowell's specific research interest lie in the areas of sonar performance analysis. Dr. S. S. Iyengar is the Chairman and Roy Paul Daniels Chaired Professor of Computer Science at Louisiana State University and is also Satish Dhawan Chaired Professor at Indian Institute of Science. He has been involved with research in high-performance algorithms, data structures, sensor fusion, data mining, and intelligent systems since receiving his Ph.D. degree (1974) and his M.S. from the Indian Institute of Science (1970). He has been a consultant to several industrial and government organizations (JPL, NASA etc.). In 1999, Professor Iyengar won the most prestigious research award titled Distinguished Research Award and a university medal for his research contributions in optimal algorithms for sensor fusion/image processing. Dr. Jianhua Chen received her Ph.D. in computer science in 1988 from Jilin University, Chang Chun, China. In August 1988, She joined the Computer Science Department of Louisiana State University, Baton Rouge, USA, where she is currently an associate professor. Dr. Chen's research interests include Machine Learning and Data Mining, Fuzzy Sets and Systems, Knowledge Representation and Reasoning.     

Privacy-preserving SVM classification

Traditional Data Mining and Knowledge Discovery algorithms assume free access to data, either at a centralized location or in federated form. Increasingly, privacy and security concerns restrict this access, thus derailing data mining projects. What is required is distributed knowledge discovery that is sensitive to this problem. The key is to obtain valid results, while providing guarantees on the nondisclosure of data. Support vector machine classification is one of the most widely used classification methodologies in data mining and machine learning. It is based on solid theoretical foundations and has wide practical application. This paper proposes a privacy-preserving solution for support vector machine (SVM) classification, PP-SVM for short. Our solution constructs the global SVM classification model from data distributed at multiple parties, without disclosing the data of each party to others. Solutions are sketched out for data that is vertically, horizontally, or even arbitrarily partitioned. We quantify the security and efficiency of the proposed method, and highlight future challenges. Jaideep Vaidya received the Bachelor’s degree in Computer Engineering from the University of Mumbai. He received the Master’s and the Ph.D. degrees in Computer Science from Purdue University. He is an Assistant Professor in the Management Science and Information Systems Department at Rutgers University. His research interests include data mining and analysis, information security, and privacy. He has received best paper awards for papers in ICDE and SIDKDD. He is a Member of the IEEE Computer Society and the ACM. Hwanjo Yu received the Ph.D. degree in Computer Science in 2004 from the University of Illinois at Urbana-Champaign. He is an Assistant Professor in the Department of Computer Science at the University of Iowa. His research interests include data mining, machine learning, database, and information systems. He is an Associate Editor of Neurocomputing and served on the NSF Panel in 2006. He has served on the program committees of 2005 ACM SAC on Data Mining track, 2005 and 2006 IEEE ICDM, 2006 ACM CIKM, and 2006 SIAM Data Mining. Xiaoqian Jiang received the B.S. degree in Computer Science from Shanghai Maritime University, Shanghai, 2003. He received the M.C.S. degree in Computer Science from the University of Iowa, Iowa City, 2005. Currently, he is pursuing a Ph.D. degree from the School of Computer Science, Carnegie Mellon University. His research interests are computer vision, machine learning, data mining, and privacy protection technologies.     

Learning with globally predictive tests

We introduce a new bias for rule learning systems. The bias only allows a rule learner to create a rule that predicts class membership if each test of the rule in isolation is predictive of that class. Although the primary motivation for the bias is to improve the understandability of rules, we show that it also improves the accuracy of learned models on a number of problems. We also introduce a related preference bias that allows creating rules that violate this restriction if they are statistically significantly better than alternative rules without such violations. Michael J. Pazzani, Ph.D.: He is a Full Professor and Chair in the Department of Information and Computer Science at the University of California, Irvine. He obtained his bachelors degree from the University of Connecticut in 1980 and his Ph. D. from University of California, Los Angles in 1987. His research interests are in machine learning, cognitive modeling and information access. He has published over 100 research papers and 2 books. He has served on the Editorial Board of the Machine Learning and the Journal of Artificial Intelligence Research.     

An instance-based learning approach based on grey relational structure

In instance-based learning, the ‘nearness’ between two instances—used for pattern classification—is generally determined by some similarity functions, such as the Euclidean or Value Difference Metric (VDM). However, Euclidean-like similarity functions are normally only suitable for domains with numeric attributes. The VDM metrics are mainly applicable to domains with symbolic attributes, and their complexity increases with the number of classes in a specific application domain. This paper proposes an instance-based learning approach to alleviate these shortcomings. Grey relational analysis is used to precisely describe the entire relational structure of all instances in a specific domain. By using the grey relational structure, new instances can be classified with high accuracy. Moreover, the total number of classes in a specific domain does not affect the complexity of the proposed approach. Forty classification problems are used for performance comparison. Experimental results show that the proposed approach yields higher performance over other methods that adopt one of the above similarity functions or both. Meanwhile, the proposed method can yield higher performance, compared to some other classification algorithms. Chi-Chun Huang is currently Assistant Professor in the Department of Information Management at National Kaohsiung Marine University, Kaohsiung, Taiwan. He received the Ph.D. degree from the Department of Electronic Engineering at National Taiwan University of Science and Technology in 2003. His research includes intelligent Internet systems, grey theory, machine learning, neural networks and pattern recognition. Hahn-Ming Lee is currently Professor in the Department of Computer Science and Information Engineering at National Taiwan University of Science and Technology, Taipei, Taiwan. He received the B.S. degree and Ph.D. degree from the Department of Computer Science and Information Engineering at National Taiwan University in 1984 and 1991, respectively. His research interests include, intelligent Internet systems, fuzzy computing, neural networks and machine learning. He is a member of IEEE, TAAI, CFSA and IICM.     

Inductive logic programming for relational knowledge discovery

Inductive logic programming (ILP) is concerned with the induction of logic programs from examples and background knowledge. In ILP, the shift of attention from program synthesis to knowledge discovery resulted in advanced techniques that are practically applicable for discovering knowledge in relational databases. This paper gives a brief introduction to ILP, presents selected ILP techniques for relational knowledge discovery and reviews selected ILP applications. Nada Lavrač, Ph.D.: She is a senior research associate at the Department of Intelligent Systems, J. Stefan Institute, Ljubljana, Slovenia (since 1978) and a visiting professor at the Klagenfurt University, Austria (since 1987). Her main research interest is in machine learning, in particular inductive logic programming and intelligent data analysis in medicine. She received a BSc in Technical Mathematics and MSc in Computer Science from Ljubljana University, and a PhD in Technical Sciences from Maribor University, Slovenia. She is coauthor of KARDIO: A Study in Deep and Qualitative Knowledge for Expert Systems, The MIT Press 1989, and Inductive Logic Programming: Techniques and Applications, Ellis Horwood 1994, and coeditor of Intelligent Data Analysis in Medicine and Pharmacology, Kluwer 1997. She was the coordinator of the European Scientific Network in Inductive Logic Programming ILPNET (1993–1996) and program cochair of the 8th European Machine Learning Conference ECML’95, and 7th International Workshop on Inductive Logic Programming ILP’97. Sašo Džeroski, Ph.D.: He is a research associate at the Department of Intelligent Systems, J. Stefan Institute, Ljubljana, Slovenia (since 1989). He has held visiting researcher positions at the Turing Institute, Glasgow (UK), Katholieke Universiteit Leuven (Belgium), German National Research Center for Computer Science (GMD), Sankt Augustin (Germany) and the Foundation for Research and Technology-Hellas (FORTH), Heraklion (Greece). His research interest is in machine learning and knowledge discovery in databases, in particular inductive logic programming and its applications and knowledge discovery in environmental databases. He is co-author of Inductive Logic Programming: Techniques and Applications, Ellis Horwood 1994. He is the scientific coordinator of ILPnet2, The Network of Excellence in Inductive Logic Programming. He was program co-chair of the 7th International Workshop on Inductive Logic Programming ILP’97 and will be program co-chair of the 16th International Conference on Machine Learning ICML’99. Masayuki Numao, Ph.D.: He is an associate professor at the Department of Computer Science, Tokyo Institute of Technology. He received a bachelor of engineering in electrical and electronics engineering in 1982 and his Ph.D. in computer science in 1987 from Tokyo Institute of Technology. He was a visiting scholar at CSLI, Stanford University from 1989 to 1990. His research interests include Artificial Intelligence, Global Intelligence and Machine Learning. Numao is a member of Information Processing Society of Japan, Japanese Society for Artificial Intelligence, Japanese Cognitive Science Society, Japan Society for Software Science and Technology and AAAI.     

Proving termination of GHC programs

A transformational approach for proving termination of parallel logic programs such as GHC programs is proposed. A transformation from GHC programs to term rewriting systems is developed; it exploits the fact that unifications in GHC-resolution correspond to matchings. The termination of a GHC program for a class of queries is implied by the termination of the resulting rewrite system. This approach facilitates the applicability of a wide range of termination techniques developed for rewrite systems in proving termination of GHC programs. The method consists of three steps: (a) deriving moding information from a given GHC program, (b) transforming the GHC program into a term rewriting system using the moding information, and finally (c) proving termination of the resulting rewrite system. Using this method, the termination of many benchmark GHC programs such as quick-sort, merge-sort, merge, split, fair-split and append, etc., can be proved. This is a revised and extended version of Ref. 12). The work was partially supported by the NSF Indo-US grant INT-9416687 Kapur was partially supported by NSF Grant nos. CCR-8906678 and INT-9014074. M. R. K. Krishna Rao, Ph.D.: He currently works as a senior research fellow at Griffith University, Brisbane, Australia. His current interests are in the areas of logic programming, modular aspects and noncopying implementations of term rewriting, learning logic programs from examples and conuterexamples and dynamics of mental states in rational agent architectures. He received his Ph.D in computer science from Tata Institute of Fundamental Research (TIFR), Bombay in 1993 and worked at TIFR and Max Planck Institut für Informatik, Saarbrücken until January 1997. Deepak Kapur, Ph.D.: He currently works as a professor at the State University of New York at Albany. His research interests are in the areas of automated reasoning, term rewriting, constraint solving, algebraic and geometric reasoning and its applications in computer vision, symbolic computation, formal methods, specification and verification. He obtained his Ph.D. in Computer Science from MIT in 1980. He worked at General Electric Corporate Research and Development until 1987. Prof. Kapur is the editor-in-chief of the Journal of Automated Reasoning. He also serves on the editorial boards of Journal of Logic Programming, Journal on Constraints, and Journal of Applicable Algebra in Engineering, Communication and Computer Science. R. K. Shyamasundar, Ph.D.: He currently works as a professor at Tata Institute of Fundamental Research (TIFR), Bombay. His current intersts are in the areas of logic programming, reactive and real time programming, constraint solving, formal methods, specification and verification. He received his Ph.D in computer science from Indian Institute of Science, Bangalore in 1975 and has been a faculty member at Tata Institute of Fundamental Research since then. He has been a visiting/regular faculty member at Technological University of Eindhoven, University of Utrecht, IBM TJ Watson Research Centre, Pennsylvania State University, University of Illinois at Urbana-Champaign, INRIA and ENSMP, France. He has served on (and chaired) Program Committees of many International Conferences and has been on the Editorial Committees.     

Reference chromosome to overcome user fatigue in IEC

Evolutionary Computation encompasses computational models that follow a biological evolution metaphor. The success of these techniques is based on the maintenance of the genetic diversity, for which it is necessary to work with large populations. However, it is not always possible to deal with such large populations, for instance, when the adequacy values must be estimated by a human being (Interactive Evolutionary Computation, IEC). This work introduces a new algorithm which is able to perform very well with a very low number of individuals (micropopulations) which speeds up the convergence and it is solving problems with complex evaluation functions. The new algorithm is compared with the canonical genetic algorithm in order to validate its efficiency. Two experimental frameworks have been chosen: table and logotype designs. An objective evaluation measures has been proposed to avoid user interaction in the experiments. In both cases the results show the efficiency of the new algorithm in terms of quality of solutions and convergence speed, two key issues in decreasing user fatigue. Yago Saez: He received the Computer Engineering degree from the Universidad Pontificia de Salamanca in 1999 Spain. He now is a Ph.D. student and works as assistant professor at the EVANNAI Group at the Computer Science Department of CARLOS III, Madrid, Spain. His main research areas encompasses the interactive evolutionary computation, the design applications and the optimization problems. Pedro Isasi, Ph.D.: He received Computer Science degree and Ph.D. degree from the Universidad Politécnica de Madrid (UPM), Spain in 1994. He is now working as professor at the EVANNAI Group at the Computer Science Department of CARLOS III, Madrid, Spain. His main research areas are Machine Learning, Evolutionary, Computation and Neural Networks and Applications to Optimization Problems. Javier Segovia, Ph.D.: He is a receiving physicist, Ph.D. degree in Computer Science (with honours) from the Universidad Politécnica de Madrid (UPM). Currently Dean of the UPM School of Computer Science, and is editor and/or author of more than 70 scientific publications in the fields of genetic algorithms, data and web mining, artificial intelligence and intelligent interfaces. Julio C. Hernandez, Ph.D.: He has received degree in Maths, Ph.D. degree in Computer Science. His main research area is the artificial intelligence applied to criptography and net security. His unofficial hobbies are chess and go. Currently, he is working as invited researcher at INRIA, France.     

Scaling Up Inductive Learning with Massive Parallelism

Machine learning programs need to scale up to very large data sets for several reasons, including increasing accuracy and discovering infrequent special cases. Current inductive learners perform well with hundreds or thousands of training examples, but in some cases, up to a million or more examples may be necessary to learn important special cases with confidence. These tasks are infeasible for current learning programs running on sequential machines. We discuss the need for very large data sets and prior efforts to scale up machine learning methods. This discussion motivates a strategy that exploits the inherent parallelism present in many learning algorithms. We describe a parallel implementation of one inductive learning program on the CM-2 Connection Machine, show that it scales up to millions of examples, and show that it uncovers special-case rules that sequential learning programs, running on smaller datasets, would miss. The parallel version of the learning program is preferable to the sequential version for example sets larger than about 10K examples. When learning from a public-health database consisting of 3.5 million examples, the parallel rule-learning system uncovered a surprising relationship that has led to considerable follow-up research.     

Approximation algorithm for periodic real-time tasks with workload-dependent running-time functions

This paper addresses the problem of resource allocation for distributed real-time periodic tasks, operating in environments that undergo unpredictable changes and that defy the specification of meaningful worst-case execution times. These tasks are supplied by input data originating from various environmental workload sources. Rather than using worst-case execution times (WCETs) to describe the CPU usage of the tasks, we assume here that execution profiles are given to describe the running time of the tasks in terms of the size of the input data of each workload source. The objective of resource allocation is to produce an initial allocation that is robust against fluctuations in the environmental parameters. We try to maximize the input size (workload) that can be handled by the system, and hence to delay possible (costly) reallocations as long as possible. We present an approximation algorithm based on first-fit and binary search that we call FFBS. As we show here, the first-fit algorithm produces solutions that are often close to optimal. In particular, we show analytically that FFBS is guaranteed to produce a solution that is at least 41% of optimal, asymptotically, under certain reasonable restrictions on the running times of tasks in the system. Moreover, we show that if at most 12% of the system utilization is consumed by input independent tasks (e.g., constant time tasks), then FFBS is guaranteed to produce a solution that is at least 33% of optimal, asymptotically. Moreover, we present simulations to compare FFBS approximation algorithm with a set of standard (local search) heuristics such as hill-climbing, simulated annealing, and random search. The results suggest that FFBS, in combination with other local improvement strategies, may be a reasonable approach for resource allocation in dynamic real-time systems. David Juedes is a tenured associate professor and assistant chair for computer science in the School of Electrical Engineering and Computer Science at Ohio University. Dr. Juedes received his Ph.D. in Computer Science from Iowa State University in 1994, and his main research interests are algorithm design and analysis, the theory of computation, algorithms for real-time systems, and bioinformatics. Dr. Juedes has published numerous conference and journal papers and has acted as a referee for IEEE Transactions on Computers, Algorithmica, SIAM Journal on Computing, Theoretical Computer Science, Information and Computation, Information Processing Letters, and other conferences and journals. Dazhang Gu is a software architect and researcher at Pegasus Technologies (NeuCo), Inc. He received his Ph.D. in Electrical Engineering and Computer Science from Ohio University in 2005. His main research interests are real-time systems, distributed systems, and resource optimization. He has published conference and journal papers on these subjects and has refereed for the Journal of Real-Time Systems, IEEE Transactions on Computers, and IEEE Transactions on Parallel and Distributed Systems among others. He also served as a session chair and publications chair for several conferences. Frank Drews is an Assistant Professor of Electical Engineering and Computer Science at Ohio Unversity. Dr. Drews received his Ph.D. in Computer Science from the Clausthal Unversity of Technolgy in Germany in 2002. His main research interests are resource management for operating systems and real-time systems, and bioinformatics. Dr. Drews has numerous publications in conferences and journals and has served as a reviewer for IEEE Transactions on Computers, the Journal of Systems and Software, and other conferences and Journals. He was Publication Chair for the OCCBIO’06 conference, Guest Editor of a Special Issue of the Journal of Systems and Software on “Dynamic Resource Management for Distributed Real-Time Systems”, organizer of special tracks at the IEEE IPDPS WPDRTS workshops in 2005 and 2006. Klaus Ecker received his Ph.D. in Theoretical Physics from the University of Graz, Austria, and his Dr. habil. in Computer Science from the University of Bonn. Since 1978 he is professor in the Department of Computer Science at the Clausthal University of Technology, Germany, and since 2005 he is visiting professor at the Ohio University. His research interests are parallel processing and theory of scheduling, especially in real time systems, and bioinformatics. Prof. Ecker published widely in the above mentioned areas in well reputed journals and proceedings of international conferences as well. He is also the author of two monographs on scheduling theory. Since 1981 he is organizing annually international workshops on parallel processing. He is associate editor of Real Time Systems, and member of the German Gesellschaft fuer Informatik (GI) and of the Association for Computing Machinery (ACM). Lonnie R. Welch received a Ph.D. in Computer and Information Science from the Ohio State University. Currently, he is the Stuckey Professor of Electrical Engineering and Computer Science at Ohio University. Dr. Welch performs research in the areas of real-time systems, distributed computing and bioinformatics. His research has been sponsored by the Defense Advanced Research Projects Agency, the Navy, NASA, the National Science Foundation and the Army. Dr. Welch has twenty years of research experience in the area of high performance computing. In his graduate work at Ohio State University, he developed a high performance 3-D graphics rendering algorithm, and he invented a parallel virtual machine for object-oriented software. For the past 15 years his research has focused on middleware and optimization algorithms for high performance computing. His research has produced three successive generations of adaptive resource management (RM) middleware for high performance real-time systems. The project has resulted in two patents and more than 150 publications. Professor Welch also collaborates on diabetes research with faculty at Edison Biotechnology Institute and on genomics research with faculty in the Department of Environmental and Plant Biology at Ohio University. Dr. Welch is a member of the editorial boards of IEEE Transactions on Computers, The Journal of Scalable Computing: Practice and Experience, and The International Journal of Computers and Applications. He is also the founder of the International Workshop on Parallel and Distributed Real-time Systems and of the Ohio Collaborative Conference on Bioinformatics. Silke Schomann graduated in 2003 with a M.Sc. in Computer Science from Clausthal University Of Technology, where she has been working as a scientific assistant since then. She is currently working on her Ph.D. thesis in computer science at the same university.     

Fault-tolerant atomic computations in an object-based distributed system

A distributed system can support fault-tolerant applications by replicating data and computation at nodes that have independent failure modes. We present a scheme called parallel execution threads (PET) which can be used to implement fault-tolerant computations in an object-based distributed system. In a system that replicates objects, the PET scheme can be used to replicate a computation by creating a number of parallel threads which execute with different replicas of the invoked objects. A computation can be completed successfully if at least one thread does not encounter any failed nodes and its completion preserves the consistency of the objects. The PET scheme can tolerate failures that occur during the execution of the computation as long as all threads are not affected by the failures. We present the algorithms required to implement the PET scheme and also address some performance issues. Mustaque Ahamad received his B.E. (Hons.) degree in Electrical Engineering from the Birla Institute of Technology and Science, Pilani, India. He obtained his M.S. and Ph.D. degrees in Computer Science from the State University of New York at Stony Brook in 1983 and 1985 respectively. Since September 1985, he is an Assistant Professor in the School of Information and Computer Science at the Georgia Institute of Technology, Atlanta. His research interests include distributed operating systems, distributed algorithms, faulttolerant systems and performance evaluation. Partha Dasgupta is an Assistant Professor at Georgia Tech since 1984. He has a Ph.D. in Computer Science from the State University of New York at Stony Brook. He is the technical project director of the Clouds distributed operating systems project, as well as a coprincipal investigator of Georgia Tech's NSF-CER award. His research interests include building distributed operating systems, distributed algorithms, fault-tolerant systems and distributed programming support. Richard J. LeBlanc, Jr. received the B.S. degree in physics from Louisiana State University in 1972 and the M.S. and Ph.D. degrees in computer sciences from the University of Wisconsin-Madison in 1974 and 1977, respectively. He is currently a Professor in the School of Information and Computer Science of the Georgia Institute of Technology. His research interests include programming language design and implementation, programming environments, and software engineering. Dr. LeBlanc's current research work involves application of these interests in distributed processing systems. As co-director of the Clouds Project, he is studying language concepts and software engineering methodology for utilizing a highly reliable, object-based distributed system. He is also interested in specification-based software development methodologies and tools. Dr. LeBlanc is a member of the Association for Computing Machinery, the IEEE Computer Society and Sigma Xi.This work was supported in part by NSF grants CCR-8619886 and CCR-8806358, and RADC contract number F30602-86-C-0032     

Equilibrium of E-markets populated with reputation oriented learning agents

In this paper, we propose reputation oriented reinforcement learning algorithms for buying and selling agents in electronic marketplaces. We consider the fact that the quality of a good offered by multiple selling agents may not be the same, and a selling agent may alter the quality of its goods. In our approach, buying agents learn to avoid the risk of purchasing low quality goods and to maximize their expected value of goods by dynamically maintaining sets of reputable and disreputable sellers. Selling agents learn to maximize their expected profits by adjusting product prices and optionally altering the quality of their goods. This paper focusses on presenting results from experiments investigating the behaviour of an e-market populated with our buying and selling agents. Our results show that such a market can reach an equilibrium state where the agent population remains stable, and this equilibrium is optimal for the participant agents. Thomas Tran, Ph.D.: He is an Assistant Professor in the School of Information Technology and Engineering at the University of Ottawa. He received his Ph.D. from the University of Waterloo in 2004. His current research work is on Multi-Agent Systems, Intelligent Agents, Reinforcement Learning, Trust and Reputation Modelling, Agent Negotiation, Mechanism Design and Applications of AI to E-Commerce. Robin Cohen, Ph.D.: She is a Professor in the School of Computer Science at the University of Waterloo. She received her Ph.D. from the University of Toronto in 1983. Her current research work is on User Modeling, Intelligent Interaction, Multi-Agent Systems, Adjustable Autonomy and Mixed-Initiative Systems and Dialogue, including Applications to E-Commerce.