Support Vector Machines for Pattern Classification

ISBN-13: 9781849960977 / Angielski / Twarda / 2010 / 473 str.

Abe Shigeo

Opis

A guide on the use of SVMs in pattern classification, including a rigorous performance comparison of classifiers and regressors. The book presents architectures for multiclass classification and function approximation problems, as well as evaluation criteria for classifiers and regressors. Features: Clarifies the characteristics of two-class SVMs; Discusses kernel methods for improving the generalization ability of neural networks and fuzzy systems; Contains ample illustrations and examples; Includes performance evaluation using publicly available data sets; Examines Mahalanobis kernels, empirical feature space, and the effect of model selection by cross-validation; Covers sparse SVMs, learning using privileged information, semi-supervised learning, multiple classifier systems, and multiple kernel learning; Explores incremental training based batch training and active-set training methods, and decomposition techniques for linear programming SVMs; Discusses variable selection for support vector regressors.

Recenzje

From the reviews:

"This broad and deep ... book is organized around the highly significant concept of pattern recognition by support vector machines (SVMs). ... The book is praxis and application oriented but with strong theoretical backing and support. Many ... details are presented and discussed, thereby making the SVM both an easy-to-understand learning machine and a more likable data modeling (mining) tool. Shigeo Abe has produced the book that will become the standard ... . I like it and therefore highly recommend this book ... ." (Vojislav Kecman, SIAM Review, Vol. 48 (2), 2006)

Spis treści

Introduction Two-Class Support Vector Machines Multiclass Support Vector Machines Variants of Support Vector Machines Training Methods Kernel-Based Methods Feature Selection and Extraction Clustering Maximum-Margin Multilayer Neural Networks Maximum-Margin Fuzzy Classifiers Function Approximation.

Blurb

Originally formulated for two-class classification problems, support vector machines (SVMs) are now accepted as powerful tools for developing pattern classification and function approximation systems. Recent developments in kernel-based methods include kernel classifiers and regressors and their variants, advancements in generalization theory, and various feature selection and extraction methods.

Providing a unique perspective on the state of the art in SVMs, with a particular focus on classification, this thoroughly updated new edition includes a more rigorous performance comparison of classifiers and regressors. In addition to presenting various useful architectures for multiclass classification and function approximation problems, the book now also investigates evaluation criteria for classifiers and regressors.

Topics and Features:

• Clarifies the characteristics of two-class SVMs through extensive analysis
• Discusses kernel methods for improving the generalization ability of conventional neural networks and fuzzy systems
• Contains ample illustrations, examples and computer experiments to help readers understand the concepts and their usefulness
• Includes performance evaluation using publicly available two-class data sets, microarray sets, multiclass data sets, and regression data sets (NEW)
• Examines Mahalanobis kernels, empirical feature space, and the effect of model selection by cross-validation (NEW)
• Covers sparse SVMs, an approach to learning using privileged information, semi-supervised learning, multiple classifier systems, and multiple kernel learning (NEW)
• Explores incremental training based batch training and active-set training methods, together with decomposition techniques for linear programming SVMs (NEW)
• Provides a discussion on variable selection for support vector regressors (NEW)

An essential guide on the use of SVMs in pattern classification, this comprehensive resource will be of interest to researchers and postgraduate students, as well as professional developers.

Dr. Shigeo Abe is a Professor at Kobe University, Graduate School of Engineering. He is the author of the Springer titles Neural Networks and Fuzzy Systems and Pattern Classification: Neuro-fuzzy Methods and Their Comparison.