Imitating Optimal Control

2018-07-01

This lecture covered further extensions to model based rl in Berkeley’s deep rl course. It provides methodologies for training a policy using model based rl methods. In particular, it introduces the GPS (Guided Policy Search) algorithm, which is somewhat resembles imitation learning on a Model Based method, and the PLATO(???) algorithm, which uses KL divergence to make DAgger more feasible.

Read More

Rudin Real and Complex Analysis Ch7

This was fun. The 7th chapter of Rudin’s RC Analysis covers differentiation and has many fun and interesting problems. At the core, many of the concepts were similar to the ones I encountered in math 2230-2240, such as the use the Jacobian (defined differently from Hubbard) and various famous theorems like Change of Variables, the definition of the derivative, and the fundamental theorem of calculus. However, the specificity for defining derivatives (using tools such as nicely shrinking sets and absolute continuity) clear up many of the inconsistencies that arise from Lebesgue integrals.

Read More