Connecting Stochastic Optimal Control and Reinforcement Learning

TL;DR

This paper explores the relationship between stochastic optimal control and reinforcement learning, demonstrating how both can be formulated within the Markov Decision Process framework and comparing two RL algorithms for control tasks.

Contribution

It establishes a formal connection between stochastic optimal control and reinforcement learning, and applies RL algorithms to solve optimal control problems.

Findings

Reformulation of optimal control as a stochastic optimization problem

Comparison of two reinforcement learning algorithms for control

Discussion on scalability and metastability challenges

Abstract

In this paper the connection between stochastic optimal control and reinforcement learning is investigated. Our main motivation is to apply importance sampling to sampling rare events which can be reformulated as an optimal control problem. By using a parameterised approach the optimal control problem becomes a stochastic optimization problem which still raises some open questions regarding how to tackle the scalability to high-dimensional problems and how to deal with the intrinsic metastability of the system. To explore new methods we link the optimal control problem to reinforcement learning since both share the same underlying framework, namely a Markov Decision Process (MDP). For the optimal control problem we show how the MDP can be formulated. In addition we discuss how the stochastic optimal control problem can be interpreted in the framework of reinforcement learning. At the end of the article we present the application of two different reinforcement learning algorithms to the optimal control problem and a comparison of the advantages and disadvantages of the two algorithms.