1
$\begingroup$

I am going to start learning the bandit problem and algorithm, especially how to bound the regret. I found the book ``Bandit Algorithms'' but it is not easy to follow. It is based on advanced stochastic processes and measure theory in some cases. I am wondering if there are any lecture notes, or courses to start.

$\endgroup$

2 Answers 2

0
$\begingroup$

I like this very detailed blog post. https://mpatacchiola.github.io/blog/2017/08/14/dissecting-reinforcement-learning-6.html. It's equivalent to a chapter or two of an introductory text, and covers a couple other adjacent but also very important topics.

$\endgroup$
1
  • 1
    $\begingroup$ Thanks for your comment. It actually covers just a few topics and it is not theory-based. I am looking for something like a course for Ph.D. level students who will do research on Bandit. $\endgroup$
    – Amin
    Commented Oct 8, 2022 at 20:41
0
$\begingroup$

You're going to need to learn measure theory, advanced probability and stochastic processes. The book neurodynamic programming covers some stochastic convergence proofs but you need some results for martingales.

In essence theoretical RL is extremely difficult, even the bandit problem is extremely.

$\endgroup$
1
  • $\begingroup$ Thanks. I would be thankful if you can recommend some references to cover these topics. These topics are so broad and may need many books. Is there any book covering the essentials of them? $\endgroup$
    – Amin
    Commented Nov 8, 2022 at 23:43

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .