TRPO (Trust Region Policy Optimization) : In depth Research Paper Review
Trust Region Policy Optimization is a fundamental paper for people working in Deep Reinforcement Learning (along with PPO or Proximal Policy Optimization) . From training Mujoco to games, TRPO has been the goto choice. It is implemented by both Unity-ML as well as OpenAI Baselines.
At 2 minute 52 seconds, the equation on right is pi a|s and not pi s|a. (minor mistake.)
Subscribe to our channel for more such videos.
In the comment section, do mention the research paper in Deep Learning or Deep RL which you want to see up next :)
Arxiv Link to the paper: https://arxiv.org/abs/1502.05477
Do consider contributing via Patreon. It can help us in making better content.
Patreon Link : / crazymuse