Stochastic Multi-Objective Multi-Armed Bandits: Regret Definition and Algorithm

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Davoodi, Mansoor, Maghsudi, Setareh
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866915346348244992
author Davoodi, Mansoor
Maghsudi, Setareh
author_facet Davoodi, Mansoor
Maghsudi, Setareh
contents Multi-armed bandit (MAB) problems are widely applied to online optimization tasks that require balancing exploration and exploitation. In practical scenarios, these tasks often involve multiple conflicting objectives, giving rise to multi-objective multi-armed bandits (MO-MAB). Existing MO-MAB approaches predominantly rely on the Pareto regret metric introduced in \cite{drugan2013designing}. However, this metric has notable limitations, particularly in accounting for all Pareto-optimal arms simultaneously. To address these challenges, we propose a novel and comprehensive regret metric that ensures balanced performance across conflicting objectives. Additionally, we introduce the concept of \textit{Efficient Pareto-Optimal} arms, which are specifically designed for online optimization. Based on our new metric, we develop a two-phase MO-MAB algorithm that achieves sublinear regret for both Pareto-optimal and efficient Pareto-optimal arms.
format Preprint
id arxiv_https___arxiv_org_abs_2506_13125
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Stochastic Multi-Objective Multi-Armed Bandits: Regret Definition and Algorithm
Davoodi, Mansoor
Maghsudi, Setareh
Machine Learning
Data Structures and Algorithms
Multi-armed bandit (MAB) problems are widely applied to online optimization tasks that require balancing exploration and exploitation. In practical scenarios, these tasks often involve multiple conflicting objectives, giving rise to multi-objective multi-armed bandits (MO-MAB). Existing MO-MAB approaches predominantly rely on the Pareto regret metric introduced in \cite{drugan2013designing}. However, this metric has notable limitations, particularly in accounting for all Pareto-optimal arms simultaneously. To address these challenges, we propose a novel and comprehensive regret metric that ensures balanced performance across conflicting objectives. Additionally, we introduce the concept of \textit{Efficient Pareto-Optimal} arms, which are specifically designed for online optimization. Based on our new metric, we develop a two-phase MO-MAB algorithm that achieves sublinear regret for both Pareto-optimal and efficient Pareto-optimal arms.
title Stochastic Multi-Objective Multi-Armed Bandits: Regret Definition and Algorithm
topic Machine Learning
Data Structures and Algorithms
url https://arxiv.org/abs/2506.13125