on 15 August 2012
This is the most complete book so far I've read on the topic with excellent description of the algorithms and models and a thorough presentation of proof. It's orientated toward Probabilistic Graphical Models only (as the title says), which means you cannot expect to read too much about other aspect of the classical Bayesian theory (I advise Christian Robert's book: A Bayesian choice, for that purpose).
The book is graduate level and needs the reader to have solid skills in linear algebra, probabilities and statistics to make the most of it.
What I really like about this book is the fact it only focus on one topic: Graphical models and do not try to cover Machine Learning in general. The consequence of that is its thorough treatment of many aspects of graphical models which is rare in the literature. That's why I highly and warmly recommend this book.
on 28 May 2016
This is an excellent but heavy going book on probabilistic graphic models. Covers most of the useful and interesting stuff in the field. But not much insight highlighted. You will need to find your gold in the book.
relevant chapters in Pattern Recognition and Machine learning by Bishop might be an easier starter, and you might learn more insight by just reading through. Come back to this book as this has much more detailed treatment, but be warned, it is very dry.