CS textbook recommendations

[banazir]

Cross-posted from my LJ.

As a hobbit birthday present, I posted the following recommendation of

Comments

Two new machine learning books

[mapjunkie]

I have two candidates to look at. Both of these are relatively new machine learning textbooks, which relate to the so-called "statistical revolution" moving machine learning from structural techniques to more statistical approaches, as exemplified by the emergence of kernel methods (canonically SVM, but in truth so much more) online improvement and resampling procedures (Adaboost leading the pack here), and rigorous theory tying learning to the theory of computation (PAC-learnability). Tom Mitchell is working on a second edition of machine learning, covering this shift. These are good books to read in the meantime.

* "Introduction to Machine Learning" by Ethem Alpaydin

If there is a threat to Tom Mitchell's "Machine Learning" as THE introductory book, this is it. Starting with an introduction in PAC-learning, it readily takes on feature-extraction, classification, clustering, sequence learning, planning, and data mining. I had a friend read three chapters, after having read virtually nothing technical for fun in years. One of the few textbooks I've read cover to cover and enjoyed heartily.

* "The Elements of Statistical Learning: Data Mining, Inference, and Prediction" by Trevor Hastie, Robert Tibshirani, and Jerome Friedman

This is THE statistician's overview of machine learning. Not for beginners, this book contains solid depth and breadth about the best practices of the new statistical techniques.