Adaptive Correlation-Weighted Intrinsic Rewards for Reinforcement Learning

TL;DR

This paper introduces ACWI, a novel adaptive intrinsic reward scaling framework for reinforcement learning that dynamically balances exploration and exploitation, leading to improved sample efficiency and stability in sparse reward environments.

Contribution

ACWI is the first method to learn a state-dependent intrinsic reward weight online using a Beta Network, enhancing exploration without manual tuning.

Findings

ACWI outperforms fixed intrinsic reward baselines in MiniGrid environments.

ACWI improves sample efficiency and training stability.

ACWI maintains computational efficiency with minimal overhead.

Abstract

We propose ACWI (Adaptive Correlation Weighted Intrinsic), an adaptive intrinsic reward scaling framework designed to dynamically balance intrinsic and extrinsic rewards for improved exploration in sparse reward reinforcement learning. Unlike conventional approaches that rely on manually tuned scalar coefficients, which often result in unstable or suboptimal performance across tasks, ACWI learns a state dependent scaling coefficient online. Specifically, ACWI introduces a lightweight Beta Network that predicts the intrinsic reward weight directly from the agent state through an encoder based architecture. The scaling mechanism is optimized using a correlation based objective that encourages alignment between the weighted intrinsic rewards and discounted future extrinsic returns. This formulation enables task adaptive exploration incentives while preserving computational efficiency and training stability. We evaluate ACWI on a suite of sparse reward environments in MiniGrid. Experimental results demonstrate that ACWI consistently improves sample efficiency and learning stability compared to fixed intrinsic reward baselines, achieving superior performance with minimal computational overhead.