Introduction to statistical pattern recognition.
2nd ed.

*(English)*Zbl 0711.62052
Computer Science and Scientific Computation. Boston, MA: Academic Press, Inc. xiii, 591 p. £36.50 (1990).

The book gives an extensive introduction to statistical pattern recognition. The content is centered around statistical models of decision making and estimation theory regarded as fundamental to the study of statistical pattern recognition. In comparison to the first, 1972 - edition the book contains a significant portion of new material, especially pertaining to feature extraction and clustering techniques. The book is self-contained.

Chapters 2 and 3 provide all necessary material on multidimensional random variables (random vectors) along with their linear transformations. Hypothesis testing including determination of error probabilities in hypothesis testing and upper bounds on the Bayes error are included in Chapter 3. Parametric classifiers (such as the Bayes linear classifier, linear and quadratic ones) are covered in Chapter 4 (Parametric classifiers). Subsequently, essential problems of parametric and nonparametric estimation are studied in Chapters 5 and 6, respectively. They constitute a rational prerequisite for studies in nonparametric classification. A thorough discussion on nonparametric classification (K Nearest Neighbor) and error estimation is included in Chapter 7.

Successive parameter estimation (such as e.g., stochastic approximation and Bayes estimation) is studied in Chapter 8. The presentation of feature extraction and linear mapping for signal representation and classification (Chapter 9 and 10) is focussed on useful techniques such as Karhunen-Loève expansion and discriminant analysis, involving several generalized criteria. Unsupervised learning (clustering) discussed in Chapter 11 addresses parametric as well as nonparametric approaches.

The book contains a vast number of computer experiments. Each chapter contains computer projects, problems and references. The material is well organized, its exposition is lucid. A major emphasis is put on concepts and algorithms. Formal proofs are not provided; in most of the cases the reader is usually referred to carefully selected references.

Due to its contents and organization the book can serve as an introductory text to statistical pattern recognition in courses on pattern recognition. It can be useful as a reference source for practitioners working in this field.

Chapters 2 and 3 provide all necessary material on multidimensional random variables (random vectors) along with their linear transformations. Hypothesis testing including determination of error probabilities in hypothesis testing and upper bounds on the Bayes error are included in Chapter 3. Parametric classifiers (such as the Bayes linear classifier, linear and quadratic ones) are covered in Chapter 4 (Parametric classifiers). Subsequently, essential problems of parametric and nonparametric estimation are studied in Chapters 5 and 6, respectively. They constitute a rational prerequisite for studies in nonparametric classification. A thorough discussion on nonparametric classification (K Nearest Neighbor) and error estimation is included in Chapter 7.

Successive parameter estimation (such as e.g., stochastic approximation and Bayes estimation) is studied in Chapter 8. The presentation of feature extraction and linear mapping for signal representation and classification (Chapter 9 and 10) is focussed on useful techniques such as Karhunen-Loève expansion and discriminant analysis, involving several generalized criteria. Unsupervised learning (clustering) discussed in Chapter 11 addresses parametric as well as nonparametric approaches.

The book contains a vast number of computer experiments. Each chapter contains computer projects, problems and references. The material is well organized, its exposition is lucid. A major emphasis is put on concepts and algorithms. Formal proofs are not provided; in most of the cases the reader is usually referred to carefully selected references.

Due to its contents and organization the book can serve as an introductory text to statistical pattern recognition in courses on pattern recognition. It can be useful as a reference source for practitioners working in this field.

Reviewer: W.Pedrycz

##### MSC:

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

68T10 | Pattern recognition, speech recognition |

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

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