Efficient self study plan

Question

I am hoping for a bit of guidance from experienced practitioners / academics.

I want to work through the Bishop ML book, but have minimal background.

What is the fastest way to get the pre-requisites (specific books would be appreciated)?

From searching around I found this potential self-study path:

• Statistical Inference - Casella / Berger

• Probability Theory and Examples - Durrett

• Linear Algebra - Hoffman / Kunze

I checked these books out from the library, but they will take me over a year to work through thoroughly, so it does not seem to be practical.

I have searched around on the internet, but most of the advice doesn't list any specific books, just what subjects I should learn.

About me

• Graduated in an unrelated discipline many years ago

• Willing to dedicate many hours to this (I am doing this to build a background for a degree in machine learning)

• I can code pretty well due to my job

Answer 1

I'm not super familiar with Bishop, but it looks like Durrett, which deals with measure-theoretic probability, would be overkill. Plus you would probably have to add real analysis to your prerequisites. I would just work through the first 4-5 Casella chapters on probability and random variables.

You probably also don't need to go through the whole linear algebra book to get started. Just make sure you understand the basic matrix operations well enough to deal with multivariate calculus (ex. taking multivariate derivatives and computing multivariate integrals).

The second half of Casella would likely be useful, but I suspect not strictly necessary.

I think if you can handle those things you'll have most of Bishop covered; you can always pick up more as needed.