What do you think?
Rate this book
Adaptive Computation and Machine Learning
Probabilistic Graphical Models: Principles and Techniques
A general framework for constructing and using probabilistic models of complex systems that would enable a computer to use available information for making decisions.
Most tasks require a person or an automated system to reason—to reach conclusions based on available information. The framework of probabilistic graphical models, presented in this book, provides a general approach for this task. The approach is model-based, allowing interpretable models to be constructed and then manipulated by reasoning algorithms. These models can also be learned automatically from data, allowing the approach to be used in cases where manually constructing a model is difficult or even impossible. Because uncertainty is an inescapable aspect of most real-world applications, the book focuses on probabilistic models, which make the uncertainty explicit and provide models that are more faithful to reality.
Probabilistic Graphical Models discusses a variety of models, spanning Bayesian networks, undirected Markov networks, discrete and continuous models, and extensions to deal with dynamical systems and relational data. For each class of models, the text describes the three fundamental cornerstones: representation, inference, and learning, presenting both basic concepts and advanced techniques. Finally, the book considers the use of the proposed framework for causal reasoning and decision making under uncertainty. The main text in each chapter provides the detailed technical development of the key ideas. Most chapters also include boxes with additional material: skill boxes, which describe techniques; case study boxes, which discuss empirical cases related to the approach described in the text, including applications in computer vision, robotics, natural language understanding, and computational biology; and concept boxes, which present significant concepts drawn from the material in the chapter. Instructors (and readers) can group chapters in various combinations, from core topics to more technically advanced material, to suit their particular needs.
Most tasks require a person or an automated system to reason—to reach conclusions based on available information. The framework of probabilistic graphical models, presented in this book, provides a general approach for this task. The approach is model-based, allowing interpretable models to be constructed and then manipulated by reasoning algorithms. These models can also be learned automatically from data, allowing the approach to be used in cases where manually constructing a model is difficult or even impossible. Because uncertainty is an inescapable aspect of most real-world applications, the book focuses on probabilistic models, which make the uncertainty explicit and provide models that are more faithful to reality.
Probabilistic Graphical Models discusses a variety of models, spanning Bayesian networks, undirected Markov networks, discrete and continuous models, and extensions to deal with dynamical systems and relational data. For each class of models, the text describes the three fundamental cornerstones: representation, inference, and learning, presenting both basic concepts and advanced techniques. Finally, the book considers the use of the proposed framework for causal reasoning and decision making under uncertainty. The main text in each chapter provides the detailed technical development of the key ideas. Most chapters also include boxes with additional material: skill boxes, which describe techniques; case study boxes, which discuss empirical cases related to the approach described in the text, including applications in computer vision, robotics, natural language understanding, and computational biology; and concept boxes, which present significant concepts drawn from the material in the chapter. Instructors (and readers) can group chapters in various combinations, from core topics to more technically advanced material, to suit their particular needs.
1270 pages, Kindle Edition
First published August 31, 2009
91 people are currently reading
1,105 people want to read
About the author
Daphne Koller
4Ìýbooks3ÌýfollowersDaphne Koller is Professor in the Department of Computer Science at Stanford University.
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 21 of 21 reviews
July 20, 2012
This is a great book for everyone, who wants to understand probabilitstic graphical models in details,
including Bayesian/Markov Networks, inference and learning from complete/incomplete data.
For getting the most from this book I recommend attending Probabilitic Graphical Models Course by Professor Daphne Koller at Stanford University ().
I wouldn't necessarily recommend it as a very first position on Bayesian Networks.
Before reading it I had read the 'Modeling and Reasoning with Bayesian Networks' book by Professor Adnan Darwiche at UCLA Computer Science Department, and it was more intuitive read, perhaps,
because of limiting the scope of a book to discrete variables and inference/learning in Bayesian Networks only.
including Bayesian/Markov Networks, inference and learning from complete/incomplete data.
For getting the most from this book I recommend attending Probabilitic Graphical Models Course by Professor Daphne Koller at Stanford University ().
I wouldn't necessarily recommend it as a very first position on Bayesian Networks.
Before reading it I had read the 'Modeling and Reasoning with Bayesian Networks' book by Professor Adnan Darwiche at UCLA Computer Science Department, and it was more intuitive read, perhaps,
because of limiting the scope of a book to discrete variables and inference/learning in Bayesian Networks only.
December 17, 2016
I was reading the book while watching Daphne Koller's video lectures. I find that the video lectures are a lot more digestible than the book. I may be wrong but it seems to me that the book may be written for people who prefer to read mathematical proofs.
Pattern Recognition and Machine Learning (PRML) by Christopher Bishop does not go into as much detail as Probabilistic Graphical Models (PGM) on probabilistic graphical models, but I find myself cross-referencing the material that PRML does cover just to get a better understanding. Some people may prefer the more verbose academic style that PGM offers, and it definitely covers probabilistic graphical models in more depth.
Pattern Recognition and Machine Learning (PRML) by Christopher Bishop does not go into as much detail as Probabilistic Graphical Models (PGM) on probabilistic graphical models, but I find myself cross-referencing the material that PRML does cover just to get a better understanding. Some people may prefer the more verbose academic style that PGM offers, and it definitely covers probabilistic graphical models in more depth.
August 11, 2013
I've finally made it through this book! For me, the only way to accomplish this was to participate in an online course (and then watch some of the videos again).
And I thought that was tough!
And I thought that was tough!
November 1, 2018
I am deeply interested in the subject of PGMs despite the fact that the area appears to be left in the dust by newer deep learning algorithms. They have a compelling 'explanatory' capability that network methods will never have. So if you work in medicine or other areas where explanatory power outweighs predictive power, PGMs are very important. It's just too bad that this is not a very good text book.
I will probably finish working on the Coursera class on PGMs but I am simply done with this textbook.
There are many much better books out there: is one that is much better by far. Another is . It would be easy to say that these two books are "easier" because perhaps they are are of limited scope, but that would be unfair to them regardless of scope.
Before you do anything in the area of PGMs you should take a look at .
I would recommend Darwiche, and/or Sucar as source texts. Even if you sign up for Koller's popular Coursera class, you can skip the book.
I will probably finish working on the Coursera class on PGMs but I am simply done with this textbook.
There are many much better books out there: is one that is much better by far. Another is . It would be easy to say that these two books are "easier" because perhaps they are are of limited scope, but that would be unfair to them regardless of scope.
Before you do anything in the area of PGMs you should take a look at .
I would recommend Darwiche, and/or Sucar as source texts. Even if you sign up for Koller's popular Coursera class, you can skip the book.
Read
July 31, 2011The contents are no doubt one of the most comprehensive (too much?) available in market about PGM. For a beginner, however, it's difficult to grasp the overall concept and to even manouver around. What's drive me insane its the usage of nonstandardized mathematical symbol making reading almost crazy. Definitely not for the faint hearted as claimed in amazon.com.
Shelved as 'reference'
October 21, 2012Yes, I really am -reading- this book.
Read
July 28, 2018This book covers a lot of topics of Probabilistic Graphical Models. It has some disadvantages like:
- Lack of examples and figures.
- It frequently refers to shapes, formulas, and tables of previous chapters which makes reading confusing.
- A lot of important proofs were skipped as exercises.
- The text is like a normal notebook of a student it could be more beautiful and organized.
With this book, I learned a lot about probabilistic graphical models but it could be easier and more fun.
- Lack of examples and figures.
- It frequently refers to shapes, formulas, and tables of previous chapters which makes reading confusing.
- A lot of important proofs were skipped as exercises.
- The text is like a normal notebook of a student it could be more beautiful and organized.
With this book, I learned a lot about probabilistic graphical models but it could be easier and more fun.
April 25, 2018
I learned a lot, although I felt it could have been subdivided into two books - an introductory one and an advanced one - as each chapter went a bit too suddenly from 'this is what these terms mean' to using those terms to describe complicated algorithms without any (much needed) reminders of what the terms mean. Also I wish the exercises had example answers.
July 12, 2020
The best book on the subject cover all the material available
November 19, 2020
Nice book, very complete. I read it as a companion to the Koller's course in Coursera.
January 9, 2022
Great book that show a lot of intricate details, serve a good complement to Daphne's PGM course on coursera.
January 17, 2023
A very good, albeit painfully dense, tour of graphical techniques
May 28, 2013
two things first: not for beginners, and not for the faint of heart. if you really need a gentle introduction, this one's not for you.
that said, this book is mostly introductions to predominant probabilistic graphical models, with each chapter trying to elaborate deeply about PGMs. the mathematical symbols used are rather different, but i suppose it's not that hard to force yourself to understand what's being written.
and don't try to skim through the book. maybe it would be more useful to read this book one chapter at a time, conditional on the kind of PGM you're interested in at the moment.
that said, this book is mostly introductions to predominant probabilistic graphical models, with each chapter trying to elaborate deeply about PGMs. the mathematical symbols used are rather different, but i suppose it's not that hard to force yourself to understand what's being written.
and don't try to skim through the book. maybe it would be more useful to read this book one chapter at a time, conditional on the kind of PGM you're interested in at the moment.
Read
October 1, 2012I finally came to terms with the fact that this book was not likely to get much more than a desultory leafing through in the near-to-intermediate future. It just never really beckoned to me from my shelf. (I traded it back to Amazon.) I had high hopes, but I just couldn't see myself slogging through 1200 pages of this when I really just want some basic insight on PGM. Fortunately, there are other titles new to my shelf that I hope will scratch that itch. I may add this one back later on, after all the errata are caught and fixed in subsequent printings.
September 14, 2012
textbook by Stanford University professor, same author is currently teaching
a course with full online video of every classroom lecture with graded homework + tests same as Stanford students - I'm taking this course - Woo Hoo
a course with full online video of every classroom lecture with graded homework + tests same as Stanford students - I'm taking this course - Woo Hoo
October 22, 2019
This book is so much more readable than Learning Bayesian Networks and really helped me to understand some concepts from my BN course. It wasn‘t perfectly understandable all the time, but then again this is a grad level course and isn‘t exactly for the layperson.
Read
May 5, 2015semua buku MIT press itu absurd kayak gini apa cuma perasaan gw doang?
Want to read
March 28, 2011graphical models are yet another topic I should probably know something about
September 25, 2013
Read this as part of a coursera course given by Daphne Koller. Comprehensive introduction to techniques and methodology used to practically use PGMs.
Read
May 13, 2015really good and comprehensive book.
Displaying 1 - 21 of 21 reviews
Can't find what you're looking for?
Get help and learn more about the design.