Characterizing the Action-Generalization Gap in Deep Q-Learning

TL;DR

This paper investigates how deep Q-learning models generalize over actions in discrete spaces, proposing evaluation methods and analyzing the factors affecting their action generalization capabilities across different domains.

Contribution

It introduces a new evaluation approach for action generalization and characterizes the action-generalization gap in deep Q-learning across various domains.

Findings

DQN can generalize over actions in simple domains

Action generalization ability decreases as action space size increases

Action generalization accelerates learning when present

Abstract

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.