Accelerating Q-learning through Efficient Value-Sharing across Actions
For deep RL practitioners, this provides a simple, parameter-free architectural addition that improves learning efficiency and reduces value overestimation in Q-learning.
The paper introduces a mean-expansion layer that accelerates Q-learning by sharing values across actions within a state, reducing value norms and overestimation. Applied to DQN and IQN, it improves aggregate performance across 57 Atari games.
Action-values are foundational to many control algorithms such as Q-learning. Therefore learning action-values efficiently is central to reinforcement learning (RL). However, learning them can be slow, requiring many updates to move values from their initialization, typically near zero, to their true values, which may be far from zero. Moreover, action-value learning algorithms typically update each state-action pair independently, without learning shared value structure across actions within a state. In this paper, we address these inefficiencies by introducing the mean-expansion layer, which accelerates action-value learning by sharing values across actions within a state and by changing the problem from directly learning potentially large action-values to learning a lower-norm representation of them. In deep RL, this layer can be applied as a parameter-free addition to Q-network architectures without altering the underlying algorithm. Applied to deep Q-networks and implicit quantile networks, it improves aggregate performance across 57 Atari games while increasing action gaps and dramatically reducing value overestimation.