Characterizing the Action-Generalization Gap in Deep Q-Learning
This addresses the problem of efficient reinforcement learning for agents needing to generalize actions, though it is incremental in characterizing existing limitations.
The paper investigates the action generalization ability of Deep Q-Networks (DQN) in discrete action spaces, finding that DQN can achieve modest generalization in simple domains but this ability decreases as the action space grows larger.
We study the action generalization ability of deep Q-learning in discrete action spaces. Generalization is crucial for efficient reinforcement learning (RL) because it allows agents to use knowledge learned from past experiences on new tasks. But while function approximation provides deep RL agents with a natural way to generalize over state inputs, the same generalization mechanism does not apply to discrete action outputs. And yet, surprisingly, our experiments indicate that Deep Q-Networks (DQN), which use exactly this type of function approximator, are still able to achieve modest action generalization. Our main contribution is twofold: first, we propose a method of evaluating action generalization using expert knowledge of action similarity, and empirically confirm that action generalization leads to faster learning; second, we characterize the action-generalization gap (the difference in learning performance between DQN and the expert) in different domains. We find that DQN can indeed generalize over actions in several simple domains, but that its ability to do so decreases as the action space grows larger.