Does anyone know any good books and review papers about Machine Learning that are fairly uptodate? From what I understand, a lot has changed from the 90s, but most of the books I seem to find are from that period  it'd be nice to see something more recent. I am familiar with neural networks (and I'm not a big fan), so I want to learn what else is out there. asked 18 Apr '11, 19:02 Iain Dunning 
I honestly advise against using the book " The Elements of Statistical Learning" nothing against professor Hastie and Tibshirani's work but I really believe that using machine learning models without considering graphical models is meaningless. Graphical model provides a way to consider conditional probabilities and modern machine learning is based on this stochastic view of the processes. Tibshirani's book completely ignores this fact and each ML technique is treated differently, making the reader overlook the stochastic model behind processes. Here is what I propose if you want to become a world class machine learning person (I am not one but trying days and nights to become one:) 1 Watch Zoubin Ghahramani's video lectures he starts by graphical models and goes into details of bayesian models (Ghahramani graduated from MIT and was Michael Jordan's student) 2 Andrew Ng's lectures are also informative 3 Wainwright and Jordan: Monograph More advanced material on graphical models and exponential families Then you can go ahead and read Tibshirani's book and it would be a great addition to what you know answered 19 Apr '11, 03:00 Mark ♦ 
Word on the street is that The Elements of Statistical Learning: is the best read in term of theory. The PDF is available for free online. answered 18 Apr '11, 20:53 tdhopper Going by the Table of Contents, that is a greatlooking book! Thanks for that.
(18 Apr '11, 22:23)
Iain Dunning

There are really two standard textbooks in machine learning. "The Elements of Statistical Learning", which was already mentioned, and "Pattern Recognition and Machine Learning" by Christopher Bishop: http://research.microsoft.com/enus/um/people/cmbishop/prml/ Very roughly speaking, the former takes a more frequentist view, while the latter is more Bayesian. It is worth having a look at both, really. answered 19 Apr '11, 13:47 Craig Schmidt 
I started to learn about machine learning through the Netflix Prize competition. The forums are a wealth of information. http://www.netflixprize.com/community/ A good place to get machine learning tools and information is the Machine Learning Open Source Software site. http://mloss.org/software/. One of the best tools to start implementing machine learning methods is free and open source statistical computing environment R http://www.rproject.org/. They have addon packages that specifically cater to Machine Learning. Another good book for Machine Learning is Clark, Fokoue, and Zhang Principles and Theory for Data Mining and Machine Learning. answered 19 Apr '11, 08:48 larrydag 1 ♦ 
If you're looking for practice, rather than theory, take a look at any of the Apache Mahout books. answered 20 Apr '11, 04:35 Geoffrey De ... ♦ 