Sanjeev Manivannan

New RL method uses policy Hessian for faster convergence

Researchers have developed a novel second-order actor-critic method for reinforcement learning in discounted Markov Decision Processes. This approach aims to accelerate convergence by incorporating curvature information…