Internally Rewarded Reinforcement Learning

TL;DR

This paper introduces Internally Rewarded Reinforcement Learning (IRRL), where rewards are generated by an internal model, and proposes a clipped linear reward function to stabilize training and improve performance.

Contribution

The paper formally defines IRRL, analyzes the effects of reward functions, and proposes a clipped linear reward function to enhance stability and convergence.

Findings

Clipped linear reward function stabilizes training in IRRL.

Proposed method achieves faster convergence.

Method outperforms baselines across diverse tasks.

Abstract

We study a class of reinforcement learning problems where the reward signals for policy learning are generated by an internal reward model that is dependent on and jointly optimized with the policy. This interdependence between the policy and the reward model leads to an unstable learning process because reward signals from an immature reward model are noisy and impede policy learning, and conversely, an under-optimized policy impedes reward estimation learning. We call this learning setting $\textit{Internally Rewarded Reinforcement Learning}$ (IRRL) as the reward is not provided directly by the environment but $\textit{internally}$ by a reward model. In this paper, we formally formulate IRRL and present a class of problems that belong to IRRL. We theoretically derive and empirically analyze the effect of the reward function in IRRL and based on these analyses propose the clipped linear reward function. Experimental results show that the proposed reward function can consistently stabilize the training process by reducing the impact of reward noise, which leads to faster convergence and higher performance compared with baselines in diverse tasks.