Effective Exploration Based on the Structural Information Principles

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zeng, Xianghua, Peng, Hao, Li, Angsheng
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909343054561280
author Zeng, Xianghua
Peng, Hao
Li, Angsheng
author_facet Zeng, Xianghua
Peng, Hao
Li, Angsheng
contents Traditional information theory provides a valuable foundation for Reinforcement Learning, particularly through representation learning and entropy maximization for agent exploration. However, existing methods primarily concentrate on modeling the uncertainty associated with RL's random variables, neglecting the inherent structure within the state and action spaces. In this paper, we propose a novel Structural Information principles-based Effective Exploration framework, namely SI2E. Structural mutual information between two variables is defined to address the single-variable limitation in structural information, and an innovative embedding principle is presented to capture dynamics-relevant state-action representations. The SI2E analyzes value differences in the agent's policy between state-action pairs and minimizes structural entropy to derive the hierarchical state-action structure, referred to as the encoding tree. Under this tree structure, value-conditional structural entropy is defined and maximized to design an intrinsic reward mechanism that avoids redundant transitions and promotes enhanced coverage in the state-action space. Theoretical connections are established between SI2E and classical information-theoretic methodologies, highlighting our framework's rationality and advantage. Comprehensive evaluations in the MiniGrid, MetaWorld, and DeepMind Control Suite benchmarks demonstrate that SI2E significantly outperforms state-of-the-art exploration baselines regarding final performance and sample efficiency, with maximum improvements of 37.63% and 60.25%, respectively.
format Preprint
id arxiv_https___arxiv_org_abs_2410_06621
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Effective Exploration Based on the Structural Information Principles
Zeng, Xianghua
Peng, Hao
Li, Angsheng
Machine Learning
Artificial Intelligence
Traditional information theory provides a valuable foundation for Reinforcement Learning, particularly through representation learning and entropy maximization for agent exploration. However, existing methods primarily concentrate on modeling the uncertainty associated with RL's random variables, neglecting the inherent structure within the state and action spaces. In this paper, we propose a novel Structural Information principles-based Effective Exploration framework, namely SI2E. Structural mutual information between two variables is defined to address the single-variable limitation in structural information, and an innovative embedding principle is presented to capture dynamics-relevant state-action representations. The SI2E analyzes value differences in the agent's policy between state-action pairs and minimizes structural entropy to derive the hierarchical state-action structure, referred to as the encoding tree. Under this tree structure, value-conditional structural entropy is defined and maximized to design an intrinsic reward mechanism that avoids redundant transitions and promotes enhanced coverage in the state-action space. Theoretical connections are established between SI2E and classical information-theoretic methodologies, highlighting our framework's rationality and advantage. Comprehensive evaluations in the MiniGrid, MetaWorld, and DeepMind Control Suite benchmarks demonstrate that SI2E significantly outperforms state-of-the-art exploration baselines regarding final performance and sample efficiency, with maximum improvements of 37.63% and 60.25%, respectively.
title Effective Exploration Based on the Structural Information Principles
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2410.06621