Poster
Convergent Policy Optimization for Safe Reinforcement Learning
Ming Yu · Zhuoran Yang · Mladen Kolar · Zhaoran Wang
East Exhibition Hall B, C #201
Keywords: [ Non-Convex Optimization ] [ Optimization ] [ Reinforcement Learning and Planning ] [ Multi-Agent RL ]
We study the safe reinforcement learning problem with nonlinear function approximation, where policy optimization is formulated as a constrained optimization problem with both the objective and the constraint being nonconvex functions. For such a problem, we construct a sequence of surrogate convex constrained optimization problems by replacing the nonconvex functions locally with convex quadratic functions obtained from policy gradient estimators. We prove that the solutions to these surrogate problems converge to a stationary point of the original nonconvex problem. Furthermore, to extend our theoretical results, we apply our algorithm to examples of optimal control and multi-agent reinforcement learning with safety constraints.
Live content is unavailable. Log in and register to view live content