Evolution Strategies (ES) as an alternative to Reinforcement Learning (RL)

Abstract

In their paper, OpenAI explores the use of Evolution Strategies (ES) algorithms as an alternative to Reinforcement Learning techniques claiming that one of this technique’s key differentiators is its scalability. I aim to compare both approaches, and I conducted experiments using the OpenAI technique, which were found to display quick learning behaviour if provided with sufficient compute resources.

Emmanuelle R. Nunes

Senior Data Scientist

My interests include statistics, machine learning, health and social statistics.