Enhancing Efficiency of Safe Reinforcement Learning via Sample Manipulation

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Gu, Shangding, Shi, Laixi, Ding, Yuhao, Knoll, Alois, Spanos, Costas, Wierman, Adam, Jin, Ming
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866909214016798720
author Gu, Shangding
Shi, Laixi
Ding, Yuhao
Knoll, Alois
Spanos, Costas
Wierman, Adam
Jin, Ming
author_facet Gu, Shangding
Shi, Laixi
Ding, Yuhao
Knoll, Alois
Spanos, Costas
Wierman, Adam
Jin, Ming
contents Safe reinforcement learning (RL) is crucial for deploying RL agents in real-world applications, as it aims to maximize long-term rewards while satisfying safety constraints. However, safe RL often suffers from sample inefficiency, requiring extensive interactions with the environment to learn a safe policy. We propose Efficient Safe Policy Optimization (ESPO), a novel approach that enhances the efficiency of safe RL through sample manipulation. ESPO employs an optimization framework with three modes: maximizing rewards, minimizing costs, and balancing the trade-off between the two. By dynamically adjusting the sampling process based on the observed conflict between reward and safety gradients, ESPO theoretically guarantees convergence, optimization stability, and improved sample complexity bounds. Experiments on the Safety-MuJoCo and Omnisafe benchmarks demonstrate that ESPO significantly outperforms existing primal-based and primal-dual-based baselines in terms of reward maximization and constraint satisfaction. Moreover, ESPO achieves substantial gains in sample efficiency, requiring 25--29% fewer samples than baselines, and reduces training time by 21--38%.
format Preprint
id arxiv_https___arxiv_org_abs_2405_20860
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Enhancing Efficiency of Safe Reinforcement Learning via Sample Manipulation
Gu, Shangding
Shi, Laixi
Ding, Yuhao
Knoll, Alois
Spanos, Costas
Wierman, Adam
Jin, Ming
Machine Learning
Safe reinforcement learning (RL) is crucial for deploying RL agents in real-world applications, as it aims to maximize long-term rewards while satisfying safety constraints. However, safe RL often suffers from sample inefficiency, requiring extensive interactions with the environment to learn a safe policy. We propose Efficient Safe Policy Optimization (ESPO), a novel approach that enhances the efficiency of safe RL through sample manipulation. ESPO employs an optimization framework with three modes: maximizing rewards, minimizing costs, and balancing the trade-off between the two. By dynamically adjusting the sampling process based on the observed conflict between reward and safety gradients, ESPO theoretically guarantees convergence, optimization stability, and improved sample complexity bounds. Experiments on the Safety-MuJoCo and Omnisafe benchmarks demonstrate that ESPO significantly outperforms existing primal-based and primal-dual-based baselines in terms of reward maximization and constraint satisfaction. Moreover, ESPO achieves substantial gains in sample efficiency, requiring 25--29% fewer samples than baselines, and reduces training time by 21--38%.
title Enhancing Efficiency of Safe Reinforcement Learning via Sample Manipulation
topic Machine Learning
url https://arxiv.org/abs/2405.20860