alomrani / imitation-learning-project

Imitation-learning-project

Updates

Mar 19- Completed Adv-CQL in ADV_CQL.py.
Feb 28- Completed functioning of SAC in SAC.py.
Feb 22- Added custom utilities to utils.py with CNN Policies in DDPG.py and TD3.py.
Feb 18- IL code base is available. Refer to README and start programming.
Feb 16- Proposal is complete. Midterm report due March 10.

General Outline

The project aims to solve/build a single research idea by balancing its theory with empirical evaluation. We hope to begin by gaining intuition about the problem and addressing it on a simple toy task. The method can then be extended to non-trivial robot control tasks in order to compare its efficacy with baseline algorithms.

A longer list of papers is available here.

Tentative Schedule

Week	Task	Description	Completed
1	Literature Review	Brainstorm Ideas and jot down good ones	✔️
2	Literature Review	Brainstorm Ideas, Meet with prof	✔️
3	Formulate Problem	Setup the problem with potential solutions	✔️
4	Implement Toy Problem	Solve base case and gain intuition	✔️
5	Implement Toy Problem	Complete base case solution and interpret results	✔️
6	Implement Algorithm	Solve main problem	✔️
7	Implement Algorithm	Solve main problem	✔️
9	Accumulate Results	Interpret and finalize results	✔️
10	Write Report	Draft and finalize report	-
11	Wrap Project	Package code base and wrap ppt	-