Pattern Recognition and Machine Learning for Self-Study II : Unsupervised Learning

This book covers unsupervised learning, one of the fundamental paradigms of pattern recognition and machine learning, and provides an accessible introduction for readers who wish to study the subject independently. Readers with a basic knowledge of linear algebra and probability theory will be able to follow the material without difficulty.

This book is the sequel to Pattern Recognition and Machine Learning for Self-Study I, which focuses on supervised learning. However, it is completely self-contained and can be read independently without prior knowledge of Volume I.

With the rapid growth of the Internet and the increasing availability of large-scale datasets, unsupervised learning has become an essential tool for discovering latent structures in data. The book begins with a detailed introduction to Bayesian statistics, followed by discussions of two major approaches to parameter estimation: maximum likelihood estimation and Bayesian estimation. It then introduces the EM algorithm, mixture models, and hidden Markov models, explaining the fundamental concepts underlying parameter estimation. Finally, it covers clustering methods. Particular attention is given to nonparametric Bayesian models, which enable clustering without requiring the number of clusters to be specified in advance.

The book has the following features:
(1) Worked examples are used extensively, and common examples are employed throughout the book to highlight the differences among various methods.
(2) Numerous experiments using real datasets and illustrative figures are included to promote intuitive understanding.
(3) Exercises are provided at the end of each chapter, and detailed solutions are available online.
(4) Appendices supplement the main text, while ‘Coffee Break’ columns provide useful insights and practical perspectives.

Readers who have studied Volume I and worked through this book will gain a broad range of practical knowledge and skills.