- Keywords: exploration, reinforcement learning, action-value methods, soft-greedy operator, softmax, mellowmax, epsilon-greedy, suboptimality gap
- Abstract: Soft-greedy operators, namely $\varepsilon$-greedy and softmax, remain a common choice to induce a basic level of exploration for action-value methods in reinforcement learning. These operators, however, have a few critical limitations. In this work, we investigate a simple soft-greedy operator, which we call resmax, that takes actions proportionally to their suboptimality gap: the residual to the estimated maximal value. It is simple to use and ensures coverage of the state-space like $\varepsilon$-greedy, but focuses exploration more on potentially promising actions like softmax. Further, it does not concentrate probability as quickly as softmax, and so better avoids overemphasizing sub-optimal actions that appear high-valued during learning. Additionally, we prove it is a non-expansion for any fixed exploration hyperparameter, unlike the softmax policy which requires a state-action specific temperature to obtain a non-expansion (called mellowmax). We empirically validate that resmax is comparable to or outperforms $\varepsilon$-greedy and softmax across a variety of environments in tabular and deep RL.
- One-sentence Summary: We propose resmax, a new soft-greedy operator for reinforment learning, which is a non-expansion and avoids overemphasis, making it a desirable replacement for the Boltzmann softmax operator.
- Supplementary Material: zip