Implicit Q-Learning and SARSA: Liberating Policy Control from Step-Size Calibration
This work addresses a critical bottleneck in reinforcement learning for practitioners by reducing the need for manual step-size tuning, though it is incremental as it builds on existing algorithms.
The paper tackles the problem of step-size sensitivity in Q-learning and SARSA reinforcement learning algorithms by proposing implicit variants that reformulate updates as fixed-point equations, resulting in adaptive step-size adjustment and reduced sensitivity, with empirical validation showing stable performance across benchmark environments.
Q-learning and SARSA are foundational reinforcement learning algorithms whose practical success depends critically on step-size calibration. Step-sizes that are too large can cause numerical instability, while step-sizes that are too small can lead to slow progress. We propose implicit variants of Q-learning and SARSA that reformulate their iterative updates as fixed-point equations. This yields an adaptive step-size adjustment that scales inversely with feature norms, providing automatic regularization without manual tuning. Our non-asymptotic analyses demonstrate that implicit methods maintain stability over significantly broader step-size ranges. Under favorable conditions, it permits arbitrarily large step-sizes while achieving comparable convergence rates. Empirical validation across benchmark environments spanning discrete and continuous state spaces shows that implicit Q-learning and SARSA exhibit substantially reduced sensitivity to step-size selection, achieving stable performance with step-sizes that would cause standard methods to fail.