Gaussian Process Policy Optimization
This work addresses the problem of efficient policy search in reinforcement learning for robotics, though it appears incremental as it builds on Proximal Policy Optimization with a Bayesian twist.
The paper tackles policy optimization in reinforcement learning by introducing a Bayesian exploration method using Gaussian processes to estimate expected returns, achieving performance comparable to or better than existing algorithms on MuJoCo robotic locomotion environments.
We propose a novel actor-critic, model-free reinforcement learning algorithm which employs a Bayesian method of parameter space exploration to solve environments. A Gaussian process is used to learn the expected return of a policy given the policy's parameters. The system is trained by updating the parameters using gradient descent on a new surrogate loss function consisting of the Proximal Policy Optimization 'Clipped' loss function and a bonus term representing the expected improvement acquisition function given by the Gaussian process. This new method is shown to be comparable to and at times empirically outperform current algorithms on environments that simulate robotic locomotion using the MuJoCo physics engine.