Support Vector Machines for Pattern Classification provides a comprehensive resource for the use of SVM?s in pattern classification. The subject area is particularly timely with research on kernel methods increasing rapidly; this book is unique in its focus on classification methods. The characteristic SVM?s are discussed: L1-SVMs and L2-SVMs, lease squares SVMs and linear programming SVMs from both a theoretical and an experimental viewpoint.
SVMs were originally formulated for two-class problems, and an extension to multiclass systems (which are essential for practical use) is not unique. However, in its discussion of several multiclass SVM architectures and the comparison of their performance using real world data, this book provides a unique perspective that researchers and students will find invaluable.
