Neighborhood Mixup Experience Replay: Local Convex Interpolation for Improved Sample Efficiency in Continuous Control Tasks

TL;DR

Neighborhood Mixup Experience Replay (NMER) enhances sample efficiency in continuous control tasks by interpolating transitions with their closest neighbors, leading to significant improvements in reinforcement learning performance.

Contribution

This paper introduces NMER, a geometrically-grounded replay buffer that performs local convex interpolation, improving sample efficiency and policy generalization in deep reinforcement learning.

Findings

NMER improves sample efficiency by 94% with TD3.

NMER improves sample efficiency by 29% with SAC.

NMER enables better recombination of experiences for learning.

Abstract

Experience replay plays a crucial role in improving the sample efficiency of deep reinforcement learning agents. Recent advances in experience replay propose using Mixup (Zhang et al., 2018) to further improve sample efficiency via synthetic sample generation. We build upon this technique with Neighborhood Mixup Experience Replay (NMER), a geometrically-grounded replay buffer that interpolates transitions with their closest neighbors in state-action space. NMER preserves a locally linear approximation of the transition manifold by only applying Mixup between transitions with vicinal state-action features. Under NMER, a given transition's set of state action neighbors is dynamic and episode agnostic, in turn encouraging greater policy generalizability via inter-episode interpolation. We combine our approach with recent off-policy deep reinforcement learning algorithms and evaluate on continuous control environments. We observe that NMER improves sample efficiency by an average 94% (TD3) and 29% (SAC) over baseline replay buffers, enabling agents to effectively recombine previous experiences and learn from limited data.