Distributed Multiarmed Bandits

doi:10.1109/TAC.2023.3247982

Distributed Multiarmed Bandits

2023-05-01

3

PRE

AI

DOI：10.1109/TAC.2023.3247982

原文链接

原文求助

摘要

En 中文

This article studies a distributed multiarmed bandit problem with heterogeneous observations of rewards. The problem is cooperatively solved by N agents assuming each agent faces a common set of M arms yet observes only local biased rewards of the arms. The goal of each agent is to minimize the cumulative expected regret with respect to the true rewards of the arms, where the mean of each arm's true reward equals the average of the means of all agents' observed biased rewards. Each agent recursively updates its decision by utilizing the information from its neighbors. Neighbor relationships are described by a time-dependent directed graph G(t) whose vertices correspond to agents and whose arcs depict neighbor relationships. A fully distributed bandit algorithm is proposed, which couples the classical distributed averaging algorithm and the celebrated upper confidence bound bandit algorithm. It is shown that for any uniformly strongly connected sequence of G(t), the algorithm achieves guaranteed regret for each agent at the order of O(log T).

Keyword:

Robot sensing systems

Distributed algorithms

Program processors

Random variables

Manipulators

Social networking (online)

Process control

Cooperative control

machine learning

agents and autonomous systems

network analysis and control

Distributed Multiarmed Bandits

摘要

期刊

机构