Ke He

New SoftGAC method improves generative actor-critic reinforcement learning

Researchers have developed a new generative actor-critic method called SoftGAC, designed to address challenges in training expressive generative policies for reinforcement learning. SoftGAC utilizes a single stochastic …