##
**NETLAB. Algorithms for pattern recognition.**
*(English)*
Zbl 1011.68116

Advances in Pattern Recognition. London: Springer. xviii, 420 p. (2002).

The book is intended to complement the book of Christopher M. Bishop, “Neural Networks for Pattern Recognition”, Oxford University Press,1995, by providing the reader the required knowledge about using NETLAB in developing pattern recognition applications. The NETLAB software toolbox was implemented in the MATLAB environment and it was designated as a suitable programming environment to provide the needed tools for the simulation of theoretically well founded neural network and related pattern recognition algorithms.

The book aims to provide readers with the knowledge and tools to get the most out of neural networks in using NETLAB facilities in solving pattern recognition tasks. It includes software implementations of several basic pattern recognition algorithms as Probabilistic Principal Component Analysis, Generative Topographic Mapping, specialized techniques of Bayesian inference and Mixture Density Networks. The book supplies both, algorithm knowledge and practical tools for a principled approach to application development by bringing together relevant theory of how to implement models efficiently and flexibly.

A series of worked examples and illustrative demonstration programs are also supplied helping the reader to understand the algorithms and how to use them. Each chapter of the book covers a group of related pattern recognition techniques and includes examples of how the techniques can be used on practical problems. The structure of the book is briefly described in the following.

The first chapter presents a brief overview of MATLAB and introduction to NETLAB toolbox. Chapter 2 covers multivariate optimisation algorithms implemented as general purpose algorithms. In Chapter 3 there is a detailed treatment of Gaussian mixture models in modelling the probability density of data. The Generalised Linear Models are presented in chapter 4 followed by the comprehensive treatment of the multi-layer perceptron given in the next chapter. The next two chapters are focused on the Radial Basis Function network, Principal Component Analysis, Generative Topographic Mapping and topographic methods. The last three chapters are all concerned with the Bayesian perspective on data modelling and inference. The Bayesian approach to neural networks is described in Chapter 9 where the evidence procedure and sampling are NETLAB implemented. The Chapter 10 develops the theory of Gaussian processes starting from Bayesian RBF used for regression.

The book provides an excellent colletction of the most important algorithms in pattern recognition. The book can be used as a textbook for teaching undergraduate and postgraduate courses in pattern recognition but it also proves extremely worth to practitioners and researchers in neurocomputing and related areas.

The book aims to provide readers with the knowledge and tools to get the most out of neural networks in using NETLAB facilities in solving pattern recognition tasks. It includes software implementations of several basic pattern recognition algorithms as Probabilistic Principal Component Analysis, Generative Topographic Mapping, specialized techniques of Bayesian inference and Mixture Density Networks. The book supplies both, algorithm knowledge and practical tools for a principled approach to application development by bringing together relevant theory of how to implement models efficiently and flexibly.

A series of worked examples and illustrative demonstration programs are also supplied helping the reader to understand the algorithms and how to use them. Each chapter of the book covers a group of related pattern recognition techniques and includes examples of how the techniques can be used on practical problems. The structure of the book is briefly described in the following.

The first chapter presents a brief overview of MATLAB and introduction to NETLAB toolbox. Chapter 2 covers multivariate optimisation algorithms implemented as general purpose algorithms. In Chapter 3 there is a detailed treatment of Gaussian mixture models in modelling the probability density of data. The Generalised Linear Models are presented in chapter 4 followed by the comprehensive treatment of the multi-layer perceptron given in the next chapter. The next two chapters are focused on the Radial Basis Function network, Principal Component Analysis, Generative Topographic Mapping and topographic methods. The last three chapters are all concerned with the Bayesian perspective on data modelling and inference. The Bayesian approach to neural networks is described in Chapter 9 where the evidence procedure and sampling are NETLAB implemented. The Chapter 10 develops the theory of Gaussian processes starting from Bayesian RBF used for regression.

The book provides an excellent colletction of the most important algorithms in pattern recognition. The book can be used as a textbook for teaching undergraduate and postgraduate courses in pattern recognition but it also proves extremely worth to practitioners and researchers in neurocomputing and related areas.

Reviewer: Luminita State (Bucureşti)

### MSC:

68T10 | Pattern recognition, speech recognition |

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

68W05 | Nonnumerical algorithms |