Risk-Averse Total-Reward Reinforcement Learning
This addresses the need for model-free algorithms in risk-averse decision-making for applications like finance or robotics, though it is incremental as it builds on existing risk measures.
The paper tackles the problem of risk-averse reinforcement learning in undiscounted infinite-horizon settings by proposing a Q-learning algorithm for entropic risk measures, demonstrating quick and reliable convergence to optimal risk-averse value functions in tabular domains.
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.