/cma-es-reinforcement-learning

CMA-ES based high confidence policy improvement for RL

Primary LanguagePython

Run main.py to start policy generation.

Generates policies in folder "Policies", used random seeds in "Seeds" and records error for those policies which did not pass the simulation return check in "Errors" folder.

For reproducing the submitted policies, please check the comments in main.py file.

Following python libraries are required to run main.py:
csv, numpy, scipy, multiprocessing, functools, tqdm, random, cma (CMA-ES library)