Beyond Surprise: Improving Exploration Through Surprise Novelty

TL;DR

This paper introduces a novel reinforcement learning model that enhances exploration by measuring the novelty of surprise through a memory network, leading to improved performance in sparse reward environments.

Contribution

It proposes a surprise memory mechanism that captures surprise novelty, addressing limitations of existing surprise-driven exploration methods.

Findings

Enhanced exploration efficiency in sparse reward settings

Significant performance improvements in Atari games

Effective reduction of attraction to noisy observations

Abstract

We present a new computing model for intrinsic rewards in reinforcement learning that addresses the limitations of existing surprise-driven explorations. The reward is the novelty of the surprise rather than the surprise norm. We estimate the surprise novelty as retrieval errors of a memory network wherein the memory stores and reconstructs surprises. Our surprise memory (SM) augments the capability of surprise-based intrinsic motivators, maintaining the agent's interest in exciting exploration while reducing unwanted attraction to unpredictable or noisy observations. Our experiments demonstrate that the SM combined with various surprise predictors exhibits efficient exploring behaviors and significantly boosts the final performance in sparse reward environments, including Noisy-TV, navigation and challenging Atari games.