Risk-Averse Total-Reward Reinforcement Learning

TL;DR

This paper introduces a Q-learning algorithm for risk-averse total-reward MDPs that guarantees convergence and performs well in tabular domains, addressing limitations of existing model-based methods.

Contribution

It develops a model-free Q-learning approach for ERM and EVaR risk measures with proven convergence and optimality guarantees.

Findings

Algorithm converges quickly in tabular domains.

Demonstrates reliable convergence to optimal risk-averse policies.

Addresses limitations of model-based algorithms requiring full transition knowledge.

Abstract

Risk-averse total-reward Markov Decision Processes (MDPs) offer a promising framework for modeling and solving undiscounted infinite-horizon objectives. Existing model-based algorithms for risk measures like the entropic risk measure (ERM) and entropic value-at-risk (EVaR) are effective in small problems, but require full access to transition probabilities. We propose a Q-learning algorithm to compute the optimal stationary policy for total-reward ERM and EVaR objectives with strong convergence and performance guarantees. The algorithm and its optimality are made possible by ERM's dynamic consistency and elicitability. Our numerical results on tabular domains demonstrate quick and reliable convergence of the proposed Q-learning algorithm to the optimal risk-averse value function.