The ability to learn from experience is a fundamental requirement for intelligence. One of the most basic characteristics of human intelligence is that people can learn from problem solving, so that they become more adept at solving problems in a given domain as they gain experience. This book investigates how computers may be programmed so that they too can learn from experience. Specifically, the aim is to take a very general, but inefficient, problem solving system and train it on a set of problems from a given domain, so that it can transform itself into a specialized, efficient problem solver for that domain. on a knowledge-intensive Recently there has been considerable progress made learning approach, explanation-based learning (EBL), that brings us closer to this possibility. As demonstrated in this book, EBL can be used to analyze a problem solving episode in order to acquire control knowledge. Control knowledge guides the problem solver's search by indicating the best alternatives to pursue at each choice point. An EBL system can produce domain specific control knowledge by explaining why the choices made during a problem solving episode were, or were not, appropriate.

Genetic algorithms (GAs) are powerful search techniques based on principles of evolution and widely applied to solve problems in many disciplines. However, most GAs employed in practice nowadays are unable to learn genetic linkage and suffer from the linkage problem. The linkage learning genetic algorithm (LLGA) was proposed to tackle the linkage problem with several specially designed mechanisms. While the LLGA performs much better on badly scaled problems than simple GAs, it does not work well on uniformly scaled problems as other competent GAs. Therefore, we need to understand why it is so and need to know how to design a better LLGA or whether there are certain limits of such a linkage learning process. This book aims to gain better understanding of the LLGA in theory and to improve the LLGA's performance in practice. It starts with a survey of the existing genetic linkage learning techniques and describes the steps and approaches taken to tackle the research topics, including using promoters, developing the convergence time model, and adopting subchromosomes.

It is common wisdom that gathering a variety of views and inputs improves the process of decision making, and, indeed, underpins a democratic society. Dubbed "ensemble learning" by researchers in computational intelligence and machine learning, it is known to improve a decision system's robustness and accuracy. Now, fresh developments are allowing researchers to unleash the power of ensemble learning in an increasing range of real-world applications. Ensemble learning algorithms such as "boosting" and "random forest" facilitate solutions to key computational issues such as face recognition and are now being applied in areas as diverse as object tracking and bioinformatics.

Responding to a shortage of literature dedicated to the topic, this volume offers comprehensive coverage of state-of-the-art ensemble learning techniques, including the random forest skeleton tracking algorithm in the Xbox Kinect sensor, which bypasses the need for game controllers. At once a solid theoretical study and a practical guide, the volume is a windfall for researchers and practitioners alike.

"

Estimation of Distribution Algorithms: A New Tool for Evolutionary Computation is devoted to a new paradigm for evolutionary computation, named estimation of distribution algorithms (EDAs). This new class of algorithms generalizes genetic algorithms by replacing the crossover and mutation operators with learning and sampling from the probability distribution of the best individuals of the population at each iteration of the algorithm. Working in such a way, the relationships between the variables involved in the problem domain are explicitly and effectively captured and exploited. This text constitutes the first compilation and review of the techniques and applications of this new tool for performing evolutionary computation. Estimation of Distribution Algorithms: A New Tool for Evolutionary Computation is clearly divided into three parts. Part I is dedicated to the foundations of EDAs. In this part, after introducing some probabilistic graphical models - Bayesian and Gaussian networks - a review of existing EDA approaches is presented, as well as some new methods based on more flexible probabilistic graphical models. A mathematical modeling of discrete EDAs is also presented. Part II covers several applications of EDAs in some classical optimization problems: the travelling salesman problem, the job scheduling problem, and the knapsack problem. EDAs are also applied to the optimization of some well-known combinatorial and continuous functions. Part III presents the application of EDAs to solve some problems that arise in the machine learning field: feature subset selection, feature weighting in K-NN classifiers, rule induction, partial abductive inference in Bayesian networks, partitional clustering, and the search for optimal weights in artificial neural networks. Estimation of Distribution Algorithms: A New Tool for Evolutionary Computation is a useful and interesting tool for researchers working in the field of evolutionary computation and for engineers who face real-world optimization problems. This book may also be used by graduate students and researchers in computer science. ... I urge those who are interested in EDAs to study this well-crafted book today.' David E. Goldberg, University of Illinois Champaign-Urbana.

A thought-provoking look at statistical learning theory and its role in understanding human learning and inductive reasoning

A joint endeavor from leading researchers in the fields of philosophy and electrical engineering, "An Elementary Introduction to Statistical Learning Theory" is a comprehensive and accessible primer on the rapidly evolving fields of statistical pattern recognition and statistical learning theory. Explaining these areas at a level and in a way that is not often found in other books on the topic, the authors present the basic theory behind contemporary machine learning and uniquely utilize its foundations as a framework for philosophical thinking about inductive inference.

Promoting the fundamental goal of statistical learning, knowing what is achievable and what is not, this book demonstrates the value of a systematic methodology when used along with the needed techniques for evaluating the performance of a learning system. First, an introduction to machine learning is presented that includes brief discussions of applications such as image recognition, speech recognition, medical diagnostics, and statistical arbitrage. To enhance accessibility, two chapters on relevant aspects of probability theory are provided. Subsequent chapters feature coverage of topics such as the pattern recognition problem, optimal Bayes decision rule, the nearest neighbor rule, kernel rules, neural networks, support vector machines, and boosting.

Appendices throughout the book explore the relationship between the discussed material and related topics from mathematics, philosophy, psychology, and statistics, drawing insightful connections between problems in these areas and statistical learning theory. All chapters conclude with a summary section, a set of practice questions, and a reference sections that supplies historical notes and additional resources for further study.

"An Elementary Introduction to Statistical Learning Theory" is an excellent book for courses on statistical learning theory, pattern recognition, and machine learning at the upper-undergraduate and graduate levels. It also serves as an introductory reference for researchers and practitioners in the fields of engineering, computer science, philosophy, and cognitive science that would like to further their knowledge of the topic.

Two key words for mechanical engineering in the future are Micro and Intelligence. It is weIl known that the leadership in the intelligence technology is a marter of vital importance for the future status of industrial society, and thus national research projects for intelligent materials, structures and machines have started not only in advanced countries, but also in developing countries. Materials and structures which have self-sensing, diagnosis and actuating systems, are called intelligent or smart, and are of growing research interest in the world. In this situation, the IUT AM symposium on Dynamics 0/ Advanced Materials and Smart Structures was a timely one. Smart materials and structures are those equipped with sensors and actuators to achieve their designed performance in achanging environment. They have complex structural properties and mechanical responses. Many engineering problems, such as interface and edge phenomena, mechanical and electro-magnetic interaction/coupling and sensing, actuating and control techniques, arise in the development ofintelligent structures. Due to the multi-disciplinary nature ofthese problems, all ofthe classical sciences and technologies, such as applied mathematics, material science, solid and fluid mechanics, control techniques and others must be assembled and used to solve them. IUTAM weIl understands the importance ofthis emerging technology. An IUTAM symposium on Smart Structures and Structronic Systems (Chaired by U.

This book honours the outstanding contributions of Vladimir Vapnik, a rare example of a scientist for whom the following statements hold true simultaneously: his work led to the inception of a new field of research, the theory of statistical learning and empirical inference; he has lived to see the field blossom; and he is still as active as ever. He started analyzing learning algorithms in the 1960s and he invented the first version of the generalized portrait algorithm. He later developed one of the most successful methods in machine learning, the support vector machine (SVM) - more than just an algorithm, this was a new approach to learning problems, pioneering the use of functional analysis and convex optimization in machine learning. Part I of this book contains three chapters describing and witnessing some of Vladimir Vapnik's contributions to science. In the first chapter, Leon Bottou discusses the seminal paper published in 1968 by Vapnik and Chervonenkis that lay the foundations of statistical learning theory, and the second chapter is an English-language translation of that original paper. In the third chapter, Alexey Chervonenkis presents a first-hand account of the early history of SVMs and valuable insights into the first steps in the development of the SVM in the framework of the generalised portrait method. The remaining chapters, by leading scientists in domains such as statistics, theoretical computer science, and mathematics, address substantial topics in the theory and practice of statistical learning theory, including SVMs and other kernel-based methods, boosting, PAC-Bayesian theory, online and transductive learning, loss functions, learnable function classes, notions of complexity for function classes, multitask learning, and hypothesis selection. These contributions include historical and context notes, short surveys, and comments on future research directions. This book will be of interest to researchers, engineers, and graduate students engaged with all aspects of statistical learning.

This thesis discusses the privacy issues in speech-based applications such as biometric authentication, surveillance, and external speech processing services. Author Manas A. Pathak presents solutions for privacy-preserving speech processing applications such as speaker verification, speaker identification and speech recognition. The author also introduces some of the tools from cryptography and machine learning and current techniques for improving the efficiency and scalability of the presented solutions. Experiments with prototype implementations of the solutions for execution time and accuracy on standardized speech datasets are also included in the text. Using the framework proposed may now make it possible for a surveillance agency to listen for a known terrorist without being able to hear conversation from non-targeted, innocent civilians."

One of the most intriguing questions about the new computer technology that has appeared over the past few decades is whether we humans will ever be able to make computers learn. As is painfully obvious to even the most casual computer user, most current computers do not. Yet if we could devise learning techniques that enable computers to routinely improve their performance through experience, the impact would be enormous. The result would be an explosion of new computer applications that would suddenly become economically feasible (e. g. , personalized computer assistants that automatically tune themselves to the needs of individual users), and a dramatic improvement in the quality of current computer applications (e. g. , imagine an airline scheduling program that improves its scheduling method based on analyzing past delays). And while the potential economic impact of successful learning methods is sufficient reason to invest in research into machine learning, there is a second significant reason: studying machine learning helps us understand our own human learning abilities and disabilities, leading to the possibility of improved methods in education. While many open questions remain about the methods by which machines and humans might learn, significant progress has been made.

One of the most intriguing questions about the new computer technology that has appeared over the past few decades is whether we humans will ever be able to make computers learn. As is painfully obvious to even the most casual computer user, most current computers do not. Yet if we could devise learning techniques that enable computers to routinely improve their performance through experience, the impact would be enormous. The result would be an explosion of new computer applications that would suddenly become economically feasible (e. g. , personalized computer assistants that automatically tune themselves to the needs of individual users), and a dramatic improvement in the quality of current computer applications (e. g. , imagine an airline scheduling program that improves its scheduling method based on analyzing past delays). And while the potential economic impact ofsuccessful learning methods is sufficient reason to invest in research into machine learning, there is a second significant reason: studying machine learning helps us understand our own human learning abilities and disabilities, leading to the possibility of improved methods in education. While many open questions remain aboutthe methods by which machines and humans might learn, significant progress has been made.

Most machine learning research has been concerned with the development of systems that implememnt one type of inference within a single representational paradigm. Such systems, which can be called monostrategy learning systems, include those for empirical induction of decision trees or rules, explanation-based generalization, neural net learning from examples, genetic algorithm-based learning, and others. Monostrategy learning systems can be very effective and useful if learning problems to which they are applied are sufficiently narrowly defined. Many real-world applications, however, pose learning problems that go beyond the capability of monostrategy learning methods. In view of this, recent years have witnessed a growing interest in developing multistrategy systems, which integrate two or more inference types and/or paradigms within one learning system. Such multistrategy systems take advantage of the complementarity of different inference types or representational mechanisms. Therefore, they have a potential to be more versatile and more powerful than monostrategy systems. On the other hand, due to their greater complexity, their development is significantly more difficult and represents a new great challenge to the machine learning community. Multistrategy Learning contains contributions characteristic of the current research in this area.

Both the way we look at data, through a DBMS, and the nature of data we ask a DBMS to manage have drastically evolved over the last decade, moving from text to images (and to sound to a lesser extent). Visual representations are used extensively within new user interfaces. Powerful visual approaches are being experimented for data manipulation, including the investigation of three dimensional display techniques. Similarly, sophisticated data visualization techniques are dramatically improving the understanding of the information extracted from a database. On the other hand, more and more applications use images as basic data or to enhance the quality and richness of data manipulation services. Image management has opened a wide area of new research topics in image understanding and analysis. The IFIP 2.6 Working Group on Databases strongly believes that a significant mutual enrichment is possible by confronting ideas, concepts and techniques supporting the work of researcher and practitioners in the two areas of visual interfaces to DBMS and DBMS management of visual data. For this reason, IFIP 2.6 has launched a series of conferences on Visual Database Systems. The first one has been held in Tokyo, 1989. VDB-2 was held in Budapest, 1991. This conference is the third in the series. As the preceding editions, the conference addresses researchers and practitioners active or interested in user interfaces, human-computer communication, knowledge representation and management, image processing and understanding, multimedia database techniques and computer vision.

Making Robots Smarter is a book about learning robots. It treats this topic based on the idea that the integration of sensing and action is the central issue. In the first part of the book, aspects of learning in execution and control are discussed. Methods for the automatic synthesis of controllers, for active sensing, for learning to enhance assembly, and for learning sensor-based navigation are presented. Since robots are not isolated but should serve us, the second part of the book discusses learning for human-robot interaction. Methods of learning understandable concepts for assembly, monitoring, and navigation are described as well as optimizing the implementation of such understandable concepts for a robot's real-time performance. In terms of the study of embodied intelligence, Making Robots Smarter asks how skills are acquired and where capabilities of execution and control come from. Can they be learned from examples or experience? What is the role of communication in the learning procedure? Whether we name it one way or the other, the methodological challenge is that of integrating learning capabilities into robots.

PHILOSOPHY AND COGNITIVE SCIENCE: CATEGORIES, CONSCIOUSNESS, AND REASONING The individual man, since his separate existence is manifested only by ignorance and error, so far as he is anything apart from his fellows, and from what he and they are to be, is only a negation. Peirce, Some Consequences of Four Incapacities. 1868. For the second time the International Colloquium on Cognitive Science gathered at San Sebastian from May, 7-11, 1991 to discuss the following main topics: Knowledge of Categories Consciousness Reasoning and Interpretation Evolution, Biology, and Mind It is not an easy task to introduce in a few words the content of this volume. We have collected eleven invited papers presented at the Colloquium, which means the substantial part of it. Unfortunately, it has not been possible to include all the invited lectures of the meeting. Before sketching and showing the relevance of each paper, let us explain the reasons for having adopted the decision to organize each two years an international colloquium on Cognitive Science at Donostia (San Sebastian). First of all, Cognitive Science is a very active research area in the world, linking multidisciplinary efforts coming mostly from psychology, artificial intelligence, theoretical linguistics and neurobiology, and using more and more formal tools. We think that this new discipline lacks solid foundations, and in this sense philosophy, particularly knowledge theory, and logic must be called for.

The work described in this book was first presented at the Second Workshop on Genetic Programming, Theory and Practice, organized by the Center for the Study of Complex Systems at the University of Michigan, Ann Arbor, 13-15 May 2004. The goal of this workshop series is to promote the exchange of research results and ideas between those who focus on Genetic Programming (GP) theory and those who focus on the application of GP to various re- world problems. In order to facilitate these interactions, the number of talks and participants was small and the time for discussion was large. Further, participants were asked to review each other's chapters before the workshop. Those reviewer comments, as well as discussion at the workshop, are reflected in the chapters presented in this book. Additional information about the workshop, addendums to chapters, and a site for continuing discussions by participants and by others can be found at http: //cscs.umich.edu:8000/GPTP-20041. We thank all the workshop participants for making the workshop an exciting and productive three days. In particular we thank all the authors, without whose hard work and creative talents, neither the workshop nor the book would be possible. We also thank our keynote speakers Lawrence ("Dave") Davis of NuTech Solutions, Inc., Jordan Pollack of Brandeis University, and Richard Lenski of Michigan State University, who delivered three thought-provoking speeches that inspired a great deal of discussion among the participants.

Designing complex programs such as operating systems, compilers, filing systems, data base systems, etc. is an old ever lasting research area. Genetic programming is a relatively new promising and growing research area. Among other uses, it provides efficient tools to deal with hard problems by evolving creative and competitive solutions. Systems Programming is generally strewn with such hard problems. This book is devoted to reporting innovative and significant progress about the contribution of genetic programming in systems programming. The contributions of this book clearly demonstrate that genetic programming is very effective in solving hard and yet-open problems in systems programming. Followed by an introductory chapter, in the remaining contributed chapters, the reader can easily learn about systems where genetic programming can be applied successfully. These include but are not limited to, information security systems, compilers, data mining systems, stock market prediction systems, robots and automatic programming.

The book discusses the analysis, comparison and integration of computational approaches to learning and research on human learning. Learning has for some time been an issue of minor importance in the cognitive sciences. It has, however, now become one of the most active research fields in psychology, the neurosciences, and computer science (machine learning).
The aim of this book is to provide the reader with an overview of the prolific research on learning throughout the disciplines.

The book will not only provide a general overview for those who are new to the field but will also provide specialist knowledge for those who want to learn more about alternative approaches and conceptualizations of learning in other disciplines. The contributing authors are all considered as leading experts in their field and come from the fields of cognitive, computer and educational science. They provide an assessment of the state-of-the-art of research, links between the disciplines, and they highlight the critically important research issues and methodologies, thus providing a basis for future research.

This book presents a modular and expandable technique in the rapidly emerging research area of automatic configuration and selection of the best algorithm for the instance at hand. The author presents the basic model behind ISAC and then details a number of modifications and practical applications. In particular, he addresses automated feature generation, offline algorithm configuration for portfolio generation, algorithm selection, adaptive solvers, online tuning, and parallelization. The author's related thesis was honorably mentioned (runner-up) for the ACP Dissertation Award in 2014, and this book includes some expanded sections and notes on recent developments. Additionally, the techniques described in this book have been successfully applied to a number of solvers competing in the SAT and MaxSAT International Competitions, winning a total of 18 gold medals between 2011 and 2014. The book will be of interest to researchers and practitioners in artificial intelligence, in particular in the area of machine learning and constraint programming.

Ontology Learning for the Semantic Web explores techniques for applying knowledge discovery techniques to different web data sources (such as HTML documents, dictionaries, etc.), in order to support the task of engineering and maintaining ontologies. The approach of ontology learning proposed in Ontology Learning for the Semantic Web includes a number of complementary disciplines that feed in different types of unstructured and semi-structured data. This data is necessary in order to support a semi-automatic ontology engineering process.
Ontology Learning for the Semantic Web is designed for researchers and developers of semantic web applications. It also serves as an excellent supplemental reference to advanced level courses in ontologies and the semantic web.

Independent Component Analysis (ICA) is a signal-processing method to extract independent sources given only observed data that are mixtures of the unknown sources. Recently, blind source separation by ICA has received considerable attention because of its potential signal-processing applications such as speech enhancement systems, telecommunications, medical signal-processing and several data mining issues. This book presents theories and applications of ICA and includes invaluable examples of several real-world applications. Based on theories in probabilistic models, information theory and artificial neural networks, several unsupervised learning algorithms are presented that can perform ICA. The seemingly different theories such as infomax, maximum likelihood estimation, negentropy maximization, nonlinear PCA, Bussgang algorithm and cumulant-based methods are reviewed and put in an information theoretic framework to unify several lines of ICA research. An algorithm is presented that is able to blindly separate mixed signals with sub- and super-Gaussian source distributions. The learning algorithms can be extended to filter systems, which allows the separation of voices recorded in a real environment (cocktail party problem). The ICA algorithm has been successfully applied to many biomedical signal-processing problems such as the analysis of electroencephalographic data and functional magnetic resonance imaging data. ICA applied to images results in independent image components that can be used as features in pattern classification problems such as visual lip-reading and face recognition systems. The ICA algorithm can furthermore be embedded in an expectation maximization framework for unsupervised classification. Independent Component Analysis: Theory and Applications is the first book to successfully address this fairly new and generally applicable method of blind source separation. It is essential reading for researchers and practitioners with an interest in ICA.

Although good devices exist for presenting visual and auditory sensations, there has yet to be a device for presenting olfactory stimulus. Nevertheless, the area for smell presentation continues to evolve and smell presentation in multimedia is not unlikely in the future. Human Olfactory Displays and Interfaces: Odor Sensing and Presentation provides the opportunity to learn about olfactory displays and its odor reproduction. Covering the fundamental and latest research of sensors and sensing systems as well as presentation technique, this book is vital for researchers, students, and practitioners gaining knowledge in the fields of consumer electronics, communications, virtual realities, electronic instruments, and more.

To-date computers are supposed to store and exploit knowledge. At least that is one of the aims of research fields such as Artificial Intelligence and Information Systems. However, the problem is to understand what knowledge means, to find ways of representing knowledge, and to specify automated machineries that can extract useful information from stored knowledge. Knowledge is something people have in their mind, and which they can express through natural language. Knowl edge is acquired not only from books, but also from observations made during experiments; in other words, from data. Changing data into knowledge is not a straightforward task. A set of data is generally disorganized, contains useless details, although it can be incomplete. Knowledge is just the opposite: organized (e.g. laying bare dependencies, or classifications), but expressed by means of a poorer language, i.e. pervaded by imprecision or even vagueness, and assuming a level of granularity. One may say that knowledge is summarized and organized data - at least the kind of knowledge that computers can store."

The analysis of experimental data is at heart of science from its beginnings.
But it was the advent of digital computers that allowed the execution of highly non-linear and increasingly complex data analysis procedures - methods that were completely unfeasible before. Non-linear curve fitting, clustering and machine learning belong to these modern techniques which are a further step towards computational intelligence.

The goal of this book is to provide an interactive and illustrative guide to these topics. It concentrates on the road from two dimensional curve fitting to multidimensional clustering and machine learning with neural networks or support vector machines. Along the way topics like mathematical optimization or evolutionary algorithms are touched. All concepts and ideas are outlined in a clear cut manner with graphically depicted plausibility arguments and a little elementary mathematics. The major topics are extensively outlined with
exploratory examples and applications. The primary goal is to be as illustrative as possible without hiding problems and pitfalls but to address them. The character of an illustrative cookbook is complemented with specific sections that address more fundamental questions like the relation between machine learning and human intelligence. These sections may be skipped without affecting
the main road but they will open up possibly interesting insights beyond the mere data massage.

All topics are completely demonstrated with the aid of the commercial computing platform Mathematica and the Computational Intelligence Packages (CIP), a high-level function library developed with Mathematica's programming language on top of Mathematica's algorithms. CIP is open-source so the detailed code of every method is freely accessible. All examples and applications shown throughout the book may be used and customized by the reader without any
restrictions.

The target readerships are students of (computer) science and engineering as well as scientific practitioners in industry and academia who deserve an illustrative introduction to these topics. Readers with programming skills may easily port and customize the provided code.
"

This book prepares students to execute the quantitative and computational needs of the finance industry. The quantitative methods are explained in detail with examples from real financial problems like option pricing, risk management, portfolio selection, etc. Codes are provided in R programming language to execute the methods. Tables and figures, often with real data, illustrate the codes. References to related work are intended to aid the reader to pursue areas of specific interest in further detail. The comprehensive background with economic, statistical, mathematical, and computational theory strengthens the understanding. The coverage is broad, and linkages between different sections are explained. The primary audience is graduate students, while it should also be accessible to advanced undergraduates. Practitioners working in the finance industry will also benefit.