##
**Pattern recognition: a statistical approach.**
*(English)*
Zbl 0542.68071

Englewood Cliffs, New Jersey etc.: Prentice-Hall International. XIV, 448 p. $ 44.95 (1982).

From the preface: ”This book originates from lecture notes prepared for an annual summer course on statistical pattern recognition which we have taught, first at Cambridge University, and lately at Oxford University, England, over a number of years. Like the course, the book is intended to cater for the tastes of a broad spectrum of readership with varying background and motivation. At one end of the scale, the material provides comprehensive information, including algorithms and procedures, for designing effective, practical pattern recognition systems. At the other end, the material is treated in depth to satisfy the inquisitiveness of the mathematically minded student who will also find challenging problems at the end of each chapter. Last but not least, it identifies research topics for those interested in the advancement of the subject.

Although the book was conceived in the context of an intensive summer course, in its present form it is well suited as a textbook for undergraduate or postgraduate courses in pattern recognition. The only prerequisite is a prior exposure to the elements of mathematics, probability theory, and statistics. The required background is usually acquired during the first years of undergraduate courses in science and engineering.

The book addresses the problems of feature evaluation, pattern classification, performance estimation, and unsupervised learning (clustering). From the outset, we felt that there was an apparent and critical need for a systematic exposition of feature evaluation methods. This subject receives here much more attention than is usually the case. This material could be included only at the price of reduced emphasis on pattern classification methods. Within this framework, we did not attempt to provide an exhaustive covering of the field. Thus, the present volume was intended as a unified exposition of carefully selected, and representative topics.”

Contents: Chapter 1. Introduction; Chapter 2. Bayes decision theory; Chapter 3. The nearest neighbor decision rule; Chapter 4. Discriminant functions; Chapter 5. Introduction to feature selection and extraction; Chapter 6. Interclass distance measures in feature selection and extraction; Chapter 7. Probabilistic separability measures in feature selection; Chapter 8. Feature extraction methods based on probabilistic separability measures; Chapter 9. Feature extraction based on the Karhunen-Loève expansion; Chapter 10. Performance estimation; Chapter 11. Nonsupervised learning pattern classification; Appendix A. Probability density function estimation; Appendix B. Differentiation of scalar functions of matrix variable; Appendix C. Properties of entropy functions.

Although the book was conceived in the context of an intensive summer course, in its present form it is well suited as a textbook for undergraduate or postgraduate courses in pattern recognition. The only prerequisite is a prior exposure to the elements of mathematics, probability theory, and statistics. The required background is usually acquired during the first years of undergraduate courses in science and engineering.

The book addresses the problems of feature evaluation, pattern classification, performance estimation, and unsupervised learning (clustering). From the outset, we felt that there was an apparent and critical need for a systematic exposition of feature evaluation methods. This subject receives here much more attention than is usually the case. This material could be included only at the price of reduced emphasis on pattern classification methods. Within this framework, we did not attempt to provide an exhaustive covering of the field. Thus, the present volume was intended as a unified exposition of carefully selected, and representative topics.”

Contents: Chapter 1. Introduction; Chapter 2. Bayes decision theory; Chapter 3. The nearest neighbor decision rule; Chapter 4. Discriminant functions; Chapter 5. Introduction to feature selection and extraction; Chapter 6. Interclass distance measures in feature selection and extraction; Chapter 7. Probabilistic separability measures in feature selection; Chapter 8. Feature extraction methods based on probabilistic separability measures; Chapter 9. Feature extraction based on the Karhunen-Loève expansion; Chapter 10. Performance estimation; Chapter 11. Nonsupervised learning pattern classification; Appendix A. Probability density function estimation; Appendix B. Differentiation of scalar functions of matrix variable; Appendix C. Properties of entropy functions.

### MSC:

68T10 | Pattern recognition, speech recognition |

62H30 | Classification and discrimination; cluster analysis (statistical aspects) |

68-01 | Introductory exposition (textbooks, tutorial papers, etc.) pertaining to computer science |

62-01 | Introductory exposition (textbooks, tutorial papers, etc.) pertaining to statistics |