arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2602.16965cs.LG
收藏

多智能体Lipschitz老虎机

Multi-Agent Lipschitz Bandits

University of Colorado Boulder(科罗拉多大学波德穆尔分校);INRIA Paris(巴黎国家信息与自动化研究所)

Sourav Chakraborty, Amit Kiran Rege, Claire Monteleoni, Lijun Chen

AI总结:

本文研究了在连续Lipschitz结构动作空间上的去中心化多玩家随机老虎机问题,其中硬碰撞导致零奖励。研究提出了一种无需通信的策略,旨在最大化集体奖励,同时分离协调成本和学习成本。通过新颖的maxima-directed搜索识别并安排玩家到高价值区域,将问题分解为N个独立的单玩家Lipschitz老虎机。在共识模式下,得到端到端的 regrets bound,其主导学习项为~O(T^{(d+1)/(d+2)}),与单玩家Lipschitz速率匹配;前期协调成本在固定置信度下与时间无关,仅在期望 regrets 形式中为多项式对数。在额外的公共覆盖/调度假设下,还获得了无间隙~O(T^{(d+1)/(d+2)})保证。进一步推导了主导学习项的匹配下界,并将框架扩展到一般距离阈值碰撞模型。

摘要:

我们研究了在连续、Lipschitz-结构化动作空间上的去中心化多玩家随机老虎机问题,其中硬碰撞导致零奖励。我们的目标是设计一种无需通信的策略,以最大化集体奖励,同时将协调成本与学习成本分开。我们提出了一种模块化协议,首先通过新颖的maxima-directed搜索识别并安排玩家到不同的高价值区域,然后将问题分解为N个独立的单玩家Lipschitz老虎机。在共识模式下,我们得到了端到端的regret界,其主导学习项为~O(T^{(d+1)/(d+2)}),与单玩家Lipschitz速率匹配;前期协调成本在固定置信度下与时间无关,仅在期望regret形式中为多项式对数。在额外的公共覆盖/调度假设下,我们还获得了无间隙~O(T^{(d+1)/(d+2)})保证。我们进一步推导了主导学习项的匹配下界,并将框架扩展到一般距离阈值碰撞模型。

英文摘要:

We study the decentralized multi-player stochastic bandit problem over a continuous, Lipschitz-structured action space where hard collisions yield zero reward. Our objective is to design a communication-free policy that maximizes collective reward, while separating coordination costs from learning costs. We propose a modular protocol that first solves the multi-agent coordination problem by identifying and seating players on distinct, high-value regions via a novel maxima-directed search and then decouples the problem into $N$ independent single-player Lipschitz bandits. In the consensus regime, we obtain an end-to-end regret bound whose dominant learning term is \(\tilde{O}(T^{(d+1)/(d+2)})\), matching the single-player Lipschitz rate; the upfront coordination cost is horizon-independent at fixed confidence and only polylogarithmic in \(T\) in the expected-regret form. Under an additional public coverage/scheduling assumption for the epochic extension, we also obtain a gap-free \(\tilde{O}(T^{(d+1)/(d+2)})\) guarantee. We further derive a matching lower bound for the dominant learning term and extend the framework to general distance-threshold collision models.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑