A Reductions Approach to Risk-Sensitive Reinforcement Learning with Optimized Certainty Equivalents

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Wang, Kaiwen, Liang, Dawen, Kallus, Nathan, Sun, Wen
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916634394886144
author Wang, Kaiwen
Liang, Dawen
Kallus, Nathan
Sun, Wen
author_facet Wang, Kaiwen
Liang, Dawen
Kallus, Nathan
Sun, Wen
contents We study risk-sensitive RL where the goal is learn a history-dependent policy that optimizes some risk measure of cumulative rewards. We consider a family of risks called the optimized certainty equivalents (OCE), which captures important risk measures such as conditional value-at-risk (CVaR), entropic risk and Markowitz's mean-variance. In this setting, we propose two meta-algorithms: one grounded in optimism and another based on policy gradients, both of which can leverage the broad suite of risk-neutral RL algorithms in an augmented Markov Decision Process (MDP). Via a reductions approach, we leverage theory for risk-neutral RL to establish novel OCE bounds in complex, rich-observation MDPs. For the optimism-based algorithm, we prove bounds that generalize prior results in CVaR RL and that provide the first risk-sensitive bounds for exogenous block MDPs. For the gradient-based algorithm, we establish both monotone improvement and global convergence guarantees under a discrete reward assumption. Finally, we empirically show that our algorithms learn the optimal history-dependent policy in a proof-of-concept MDP, where all Markovian policies provably fail.
format Preprint
id arxiv_https___arxiv_org_abs_2403_06323
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle A Reductions Approach to Risk-Sensitive Reinforcement Learning with Optimized Certainty Equivalents
Wang, Kaiwen
Liang, Dawen
Kallus, Nathan
Sun, Wen
Machine Learning
We study risk-sensitive RL where the goal is learn a history-dependent policy that optimizes some risk measure of cumulative rewards. We consider a family of risks called the optimized certainty equivalents (OCE), which captures important risk measures such as conditional value-at-risk (CVaR), entropic risk and Markowitz's mean-variance. In this setting, we propose two meta-algorithms: one grounded in optimism and another based on policy gradients, both of which can leverage the broad suite of risk-neutral RL algorithms in an augmented Markov Decision Process (MDP). Via a reductions approach, we leverage theory for risk-neutral RL to establish novel OCE bounds in complex, rich-observation MDPs. For the optimism-based algorithm, we prove bounds that generalize prior results in CVaR RL and that provide the first risk-sensitive bounds for exogenous block MDPs. For the gradient-based algorithm, we establish both monotone improvement and global convergence guarantees under a discrete reward assumption. Finally, we empirically show that our algorithms learn the optimal history-dependent policy in a proof-of-concept MDP, where all Markovian policies provably fail.
title A Reductions Approach to Risk-Sensitive Reinforcement Learning with Optimized Certainty Equivalents
topic Machine Learning
url https://arxiv.org/abs/2403.06323