Today I watched lecture 5 from Berkeley’s deep rl course, which covered actor critic algorithms.
I covered the second chapter of Rudin’s Real and Complex Analysis. It’s still pretty straightfoward, but some of the techniques used are getting tougher to keep up with. I might start taking notes over the theorems, since I’m finding myself struggling to recall everything, which greatly hampers the learning process. Here are my solutions to the odd questions, although some of them may not be well explained.
Today I watched lecture 4 from Berkeley’s deep rl course, which covered policy gradients. Policy gradients, which heavily rely on sampled data, actually have some really nice math behind them! In particular, there was one interesting trick that helped separate out the derivative for the policy, which made it actually workable! I include the notes below:
I covered the first chapter in Rudin’s Real and Complex Analysis. I’m planning to go over this book before I take Math 6110 this fall (since I don’t technically have experience in analysis). Here are some of my first impressions