My formal goal is to be able to rigorously understand the mathematical basis for modern statistical learning methods (ML, deep learning). I am told by math people that this involves: linear algebra, multivariable calc, probability theory, statistics. My current strategy is to go through classic textbooks, one subject at a time, and do most the problems (e.g. Axler's linear algebra, Casella-Berger for stats). Is there a more efficient strategy than this? I have heard of Ravi Vakil's advice to learn about things you don't really understand, then backfill knowledge gaps, but I'm not sure what that would look like in practice, and his advice seems to be for very advanced math students.

Math Background:

I took computer science and math classes in college, but didn't have or gain mathematical maturity. Since then I read Pinter's abstract algebra and Abbott's analysis text and did most the problems. This took a really long time and a lot of willpower, hence the question.


I am a full time student in a nontechnical field that I think can benefit from AI. So I can't read math full time, and ultimately want to apply AI, but I also want to do it properly and enjoy math for its own sake.

  • $\begingroup$ I recommend also looking at Strang's books on linear algebra. $\endgroup$ – littleO Jul 25 '16 at 20:42
  • $\begingroup$ the best strategy is to prove everything with your own words (and I mean prove the linear algebra and probability theorems/algorithms used in machine learning, not the abstract concepts you'll find in math books) $\endgroup$ – reuns Jul 25 '16 at 23:44
  • $\begingroup$ statweb.stanford.edu/~tibs/ElemStatLearn $\endgroup$ – John M Jul 26 '16 at 4:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.