Efficient Reinforcement Learning Using Gaussian Processes

摘要: This book examines Gaussian processes in both model-based reinforcement learning (RL) and inference nonlinear dynamic systems. First, we introduce PILCO, a fully Bayesian approach for efficient RL continuous-valued state action spaces when no expert knowledge is available. PILCO takes model uncertainties consistently into account during long-term planning to reduce bias. Second, propose principled algorithms robust filtering smoothing GP systems.