from Hacker News

Experts recommend Machine Learning books

by dmonn on 7/30/20, 12:28 PM with 20 comments

• by henrik_w on 7/30/20, 1:59 PM

  For a survey of AI and ML I really liked "Artificial Intelligence – A Guide for Thinking Humans" by Melanie Mitchell. I've written a summary of it here: https://henrikwarne.com/2020/05/19/artificial-intelligence-a...

• by cdavid on 7/30/20, 3:14 PM

  The list is decent, but not exactly original.

  For people w/ a physics background, I would still recommend https://www.inference.org.uk/itprnn/book.pdf. Some of it is a bit obsolete, but then DL made a lot of stuff around generalization/overfitting somehow obsolete. It makes a lot of connection between different kind of approaches in ML, information theory, (Bayesian) statistics, and physics.

  It is not a very good book if you only care about applications (in which case the Keras book, for beginner, or fastai/etc. are much more appropriate).

• by melenaboija on 7/30/20, 1:34 PM

  For NLP I would maybe add

  Speech and Language Processing From Dan Jurafsky,

  Available at: https://web.stanford.edu/~jurafsky/slp3/

• by inopinatus on 7/30/20, 1:25 PM

  Slightly crestfallen that the “ML” here is machine learning and not the programming language. I still refer to my vintage paperback of L. C. Paulson’s ML for the Working Programmer from time to time.

• by newtohn99 on 7/30/20, 5:59 PM

  As a new grad (bachelors swe), is it worth it to jump on the ML hype train ? I see modelling is almost always only open to phds/masters.

  So is studying all that stuff just for being a MLE/ data engineer worth it, if you are already a software developer (full stack)?

• by horsemessiah on 7/30/20, 1:09 PM

  State and Revolution is where I'd recommend people start for ML ;)