SSRNML & AI Methods

Model-Based Reinforcement Learning in Diffusion Environments

A new framework for continuous-time model-based reinforcement learning, which reduces the discrepancy between the model-based value function and empirical rewards, could improve upon traditional statistical methods.