Read the second half in a day to prep for a machine learning course. I can’t say that I found it particularly helpful but that’s probably my fault morRead the second half in a day to prep for a machine learning course. I can’t say that I found it particularly helpful but that’s probably my fault more so than the book’s. Still, after browsing through the table of contents I think the first half would be valuable for intuition but the second was just too terse to provide good theoretical foundations. Will have to read Ross� A First Course in P Theory over the holidays to better understand this stuff. ...more
*** Second read*** The first time I read this I thought it was too terse/uncomprehensive but this second time around I found it particularly helpful. *** Second read*** The first time I read this I thought it was too terse/uncomprehensive but this second time around I found it particularly helpful. Initially the LA material is a bit too light but as you get to the Eigenstuff it improves vastly. I also really enjoyed the later chapters on linear regression (max likelihood, max a posteriori, and Bayesian approaches), on support vector machines (they have a nice derivation of the dual optimisation problem), and lastly on gaussian mixture models. Can highly recommend this as a supplement to other ML textbooks.
*** First read*** Not a fan. Didn’t like the notation and on top of that they barely proved any theorems. Felt like I was mindlessly going through a long chain of meaningless theorem. Personally think that the only way to learn this stuff properly is by doing a deep dive into linear algebra (Axler’s Linear Algebra Done Right or MIT OCW), multi-variable calculus (MIT OCW), probability theory (Ross?) and a proper ML textbook like Bishop’s Pattern Recognition and Machine Learning....more
not the best introduction to the subject but acts as a nice supplement to the books by Strang and Axler. My main issue was that the notation was inconnot the best introduction to the subject but acts as a nice supplement to the books by Strang and Axler. My main issue was that the notation was inconsistent, the numbering of the theorems and definitions weird (unlike in Axler say), and the back and forth between the matrix and linear map point of view a bit confusing. What I did really like however, were his discussions of change of basis matrices/maps, and the various points of view he gave on matrix multiplication. All in all, I’m glad I read it, but honestly think the path via Strang’s MIT course (the free one)/his textbook and Axler's Linear Algebra Done Right is the best way through the subject. If, however, you’re in a rush, and your somewhat familiar with mathematical proofs, this is a pretty good option....more
Loading...
