Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Satheesh, Anirudh, Powell, Keenan, Aggarwal, Vaneet
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2511.05694
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866918377201598464
author Satheesh, Anirudh
Powell, Keenan
Aggarwal, Vaneet
author_facet Satheesh, Anirudh
Powell, Keenan
Aggarwal, Vaneet
contents A central challenge in reinforcement learning is that policies trained in controlled environments often fail under distribution shifts at deployment into real-world environments. Distributionally Robust Reinforcement Learning (DRRL) addresses this by optimizing for worst-case performance within an uncertainty set defined by a robustness budget $ε$. However, fixing $ε$ results in a tradeoff between performance and robustness: small values yield high nominal performance but weak robustness, while large values can result in instability and overly conservative policies. We propose Distributionally Robust Self-Paced Curriculum Reinforcement Learning (DR-SPCRL), a method that overcomes this limitation by treating $ε$ as a continuous curriculum. DR-SPCRL adaptively schedules the robustness budget according to the agent's progress, enabling a balance between nominal and robust performance. Empirical results across multiple environments demonstrate that DR-SPCRL not only stabilizes training but also achieves a superior robustness-performance trade-off, yielding an average 11.8\% increase in episodic return under varying perturbations compared to fixed or heuristic scheduling strategies, and achieving approximately 1.9$\times$ the performance of the corresponding nominal RL algorithms.
format Preprint
id arxiv_https___arxiv_org_abs_2511_05694
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Distributionally Robust Self Paced Curriculum Reinforcement Learning
Satheesh, Anirudh
Powell, Keenan
Aggarwal, Vaneet
Machine Learning
A central challenge in reinforcement learning is that policies trained in controlled environments often fail under distribution shifts at deployment into real-world environments. Distributionally Robust Reinforcement Learning (DRRL) addresses this by optimizing for worst-case performance within an uncertainty set defined by a robustness budget $ε$. However, fixing $ε$ results in a tradeoff between performance and robustness: small values yield high nominal performance but weak robustness, while large values can result in instability and overly conservative policies. We propose Distributionally Robust Self-Paced Curriculum Reinforcement Learning (DR-SPCRL), a method that overcomes this limitation by treating $ε$ as a continuous curriculum. DR-SPCRL adaptively schedules the robustness budget according to the agent's progress, enabling a balance between nominal and robust performance. Empirical results across multiple environments demonstrate that DR-SPCRL not only stabilizes training but also achieves a superior robustness-performance trade-off, yielding an average 11.8\% increase in episodic return under varying perturbations compared to fixed or heuristic scheduling strategies, and achieving approximately 1.9$\times$ the performance of the corresponding nominal RL algorithms.
title Distributionally Robust Self Paced Curriculum Reinforcement Learning
topic Machine Learning
url https://arxiv.org/abs/2511.05694