Probabilistic Graphical Models
Principles and Applications
This accessible text/reference offers a comprehensive introduction to probabilistic graphical models (PGMs) from an engineering viewpoint. It covers the fundamentals of key PGM classes, including representation, inference, and learning principles, while reviewing real-world applications across various disciplines. Applications include Bayesian classifiers, hidden Markov models, Bayesian networks, dynamic and temporal Bayesian networks, Markov random fields, influence diagrams, and Markov decision processes. Key features include a unified framework for all main PGM classes, exploration of fundamental aspects of representation, inference, and learning for each technique, and practical applications of different methods. The text also examines recent developments in the field, such as multidimensional Bayesian classifiers, relational graphical models, and causal models. Each chapter concludes with exercises, further reading suggestions, and ideas for research or programming projects, alongside course outlines for instructors in the preface. This classroom-tested work serves as a textbook for advanced undergraduate or graduate courses in probabilistic graphical models for students in computer science, engineering, and physics. It is also a valuable reference for professionals looking to apply PGMs in their fields or seeking to understand the foundational techniques.
