Density estimation has evolved enormously since the days of bar plots and histograms, but researchers and users are still struggling with the problem of the selection of the bin widths. This book is the first to explore a new paradigm for the data-based or automatic selection of the free parameters of density estimates in general so that the expected error is within a given constant multiple of the best possible error. The paradigm can be used in nearly all density estimates and for most model selection problems, both parametric and nonparametric.

Density estimation has evolved enormously since the days of bar plots and histograms, but researchers and users are still struggling with the problem of the selection of the bin widths. This text explores a new paradigm for the data-based or automatic selection of the free parameters of density estimates in general so that the expected error is within a given constant multiple of the best possible error. The paradigm can be used in nearly all density estimates and for most model selection problems, both parametric and nonparametric. It is the first book on this topic. The text is intended for first-year graduate students in statistics and learning theory, and offers a host of opportunities for further research and thesis topics. Each chapter corresponds roughly to one lecture, and is supplemented with many classroom exercises. A one year course in probability theory at the level of Feller's Volume 1 should be more than adequate preparation. Gabor Lugosi is Professor at Universitat Pompeu Fabra in Barcelona, and Luc Debroye is Professor at McGill University in Montreal. In 1996, the authors, together with Lászlo Györfi, published the successful text, A Probabilistic Theory of Pattern Recognition with Springer-Verlag. Both authors have made many contributions in the area of nonparametric estimation.

Density estimation has evolved enormously since the days of bar plots and histograms, but researchers and users are still struggling with the problem of the selection of the bin widths. This book is the first to explore a new paradigm for the data-based or automatic selection of the free parameters of density estimates in general so that the expected error is within a given constant multiple of the best possible error. The paradigm can be used in nearly all density estimates and for most model selection problems, both parametric and nonparametric.

In many statistical applications, scientists have to analyze the occurrence of observed clusters of events in time or space. Scientists are especially interested in determining whether an observed cluster of events has occurred by chance if it is assumed that the events are distributed independently and uniformly over time or space. Scan statistics have relevant applications in many areas of science and technology including geology, geography, medicine, minefield detection, molecular biology, photography, quality control and reliability theory and radio-optics.

This book discusses dynamical systems that are typically driven by stochastic dynamic noise. It is written by two statisticians essentially for the statistically inclined readers. It covers many of the contributions made by the statisticians in the past twenty years or so towards our understanding of estimation, the Lyapunov-like index, the nonparametric regression, and many others, many of which are motivated by their dynamical system counterparts but have now acquired a distinct statistical flavor.

Over the last few decades, important progresses in the methods of sampling have been achieved. This book draws up an inventory of new methods that can be useful for selecting samples. Forty-six sampling methods are described in the framework of general theory. The algorithms are described rigorously, which allows implementing directly the described methods. This book is aimed at experienced statisticians who are familiar with the theory of survey sampling.

Science has made great progress in the twentieth century, with the establishment of proper disciplines in the fields of physics, computer science, molecular biology, and many others. At the same time, there have also emerged many engineering ideas that are interdisciplinary in nature, beyond the realm of such orthodox disciplines. These in clude, for example, artificial intelligence, fuzzy logic, artificial neural networks, evolutional computation, data mining, and so on. In or der to generate new technology that is truly human-friendly in the twenty-first century, integration of various methods beyond specific disciplines is required. Soft computing is a key concept for the creation of such human friendly technology in our modern information society. Professor Rutkowski is a pioneer in this field, having devoted himself for many years to publishing a large variety of original work. The present vol ume, based mostly on his own work, is a milestone in the devel opment of soft computing, integrating various disciplines from the fields of information science and engineering. The book consists of three parts, the first of which is devoted to probabilistic neural net works. Neural excitation is stochastic, so it is natural to investi gate the Bayesian properties of connectionist structures developed by Professor Rutkowski. This new approach has proven to be par ticularly useful for handling regression and classification problems vi Preface in time-varying environments. Throughout this book, major themes are selected from theoretical subjects that are tightly connected with challenging applications.

There have been major developments in the field of statistics over the last quarter century, spurred by the rapid advances in computing and data-measurement technologies. These developments have revolutionized the field and have greatly influenced research directions in theory and methodology. Increased computing power has spawned entirely new areas of research in computationally-intensive methods, allowing us to move away from narrowly applicable parametric techniques based on restrictive assumptions to much more flexible and realistic models and methods. These computational advances have also led to the extensive use of simulation and Monte Carlo techniques in statistical inference. All of these developments have, in turn, stimulated new research in theoretical statistics. This volume provides an up-to-date overview of recent advances in statistical modeling and inference. Written by renowned researchers from across the world, it discusses flexible models, semi-parametric methods and transformation models, nonparametric regression and mixture models, survival and reliability analysis, and re-sampling techniques. With its coverage of methodology and theory as well as applications, the book is an essential reference for researchers, graduate students, and practitioners.

Now available in paperback, this book provides a comprehensive account of survey sampling theory and methodology suitable for students and researchers across a variety of disciplines. It shows how statistical modeling is a vital component of the sampling process and in the choice of estimation technique. The first textbook that systematically extends traditional sampling theory with the aid of a modern model assisted outlook. Covers classical topics as well as areas where significant new developments have taken place.

This book is intended for students and practitioners who have had a calculus-based statistics course and who have an interest in safety considerations such as reliability, strength, and duration-of-load or service life. Many persons studying statistical science will be employed professionally where the problems encountered are obscure, what should be analyzed is not clear, the appropriate assumptions are equivocal, and data are scant. In this book there is no disclosure with many of the data sets what type of investigation should be made or what assumptions are to be used.