Measuring and Mitigating Interference in Reinforcement Learning

TL;DR

This paper introduces a new measure of interference in value-based reinforcement learning, demonstrating its correlation with instability, and proposes online-aware algorithms that effectively reduce interference and improve control performance.

Contribution

The work provides a novel interference measure for reinforcement learning, enabling better understanding and mitigation of interference effects in neural network-based control systems.

Findings

Interference correlates with control instability.

Online-aware algorithms reduce interference and improve stability.

The measure enables new scientific insights into deep learning architectures.

Abstract

Catastrophic interference is common in many network-based learning systems, and many proposals exist for mitigating it. Before overcoming interference we must understand it better. In this work, we provide a definition and novel measure of interference for value-based reinforcement learning methods such as Fitted Q-Iteration and DQN. We systematically evaluate our measure of interference, showing that it correlates with instability in control performance, across a variety of network architectures. Our new interference measure allows us to ask novel scientific questions about commonly used deep learning architectures and study learning algorithms which mitigate interference. Lastly, we outline a class of algorithms which we call online-aware that are designed to mitigate interference, and show they do reduce interference according to our measure and that they improve stability and performance in several classic control environments.