Distributional Hamilton-Jacobi-Bellman Equations for Continuous-Time Reinforcement Learning
This work addresses risk-sensitive control and other applications in continuous-time RL, representing an incremental advancement by extending distributional methods to continuous-time settings.
The paper tackles the problem of predicting return distributions in continuous-time reinforcement learning by establishing a distributional analogue of the Hamilton-Jacobi-Bellman equation for stochastic processes and proposing a tractable algorithm based on a JKO scheme, demonstrating its effectiveness in a synthetic control problem.
Continuous-time reinforcement learning offers an appealing formalism for describing control problems in which the passage of time is not naturally divided into discrete increments. Here we consider the problem of predicting the distribution of returns obtained by an agent interacting in a continuous-time, stochastic environment. Accurate return predictions have proven useful for determining optimal policies for risk-sensitive control, learning state representations, multiagent coordination, and more. We begin by establishing the distributional analogue of the Hamilton-Jacobi-Bellman (HJB) equation for Itô diffusions and the broader class of Feller-Dynkin processes. We then specialize this equation to the setting in which the return distribution is approximated by $N$ uniformly-weighted particles, a common design choice in distributional algorithms. Our derivation highlights additional terms due to statistical diffusivity which arise from the proper handling of distributions in the continuous-time setting. Based on this, we propose a tractable algorithm for approximately solving the distributional HJB based on a JKO scheme, which can be implemented in an online control algorithm. We demonstrate the effectiveness of such an algorithm in a synthetic control problem.