When does Self-Prediction help? Understanding Auxiliary Tasks in Reinforcement Learning

TL;DR

This paper analyzes when auxiliary tasks like self-prediction aid reinforcement learning, providing theoretical insights and empirical evidence on their effectiveness and interactions with distractions and observation functions.

Contribution

It offers a theoretical framework explaining the benefits of latent self-prediction as an auxiliary task and its interaction with observation reconstruction in reinforcement learning.

Findings

Latent self-prediction is a helpful auxiliary task.

Observation reconstruction provides useful features in isolation.

The linear model analysis predicts behaviors in non-linear neural networks.

Abstract

We investigate the impact of auxiliary learning tasks such as observation reconstruction and latent self-prediction on the representation learning problem in reinforcement learning. We also study how they interact with distractions and observation functions in the MDP. We provide a theoretical analysis of the learning dynamics of observation reconstruction, latent self-prediction, and TD learning in the presence of distractions and observation functions under linear model assumptions. With this formalization, we are able to explain why latent-self prediction is a helpful \emph{auxiliary task}, while observation reconstruction can provide more useful features when used in isolation. Our empirical analysis shows that the insights obtained from our learning dynamics framework predicts the behavior of these loss functions beyond the linear model assumption in non-linear neural networks. This reinforces the usefulness of the linear model framework not only for theoretical analysis, but also practical benefit for applied problems.