Year: 2021
Type: preprint
Abstract: This paper presents a novel federated linear contextual bandits model, where individual clients face different $K$-armed stochastic bandits coupled through common global parameters. By leveraging the geometric structure of the linear rewards, a collaborative algorithm called Fed-PE is proposed to cope with the heterogeneity across clients without exchanging local feature vectors or raw data. Fed-PE relies on a novel multi-client G-optimal design, and achieves near-optimal regrets for both disjoint and shared parameter cases with logarithmic communication costs. In addition, a new concept called collinearly-dependent policies is introduced, based on which a tight minimax regret lower bound for the disjoint parameter case is derived. Experiments demonstrate the effectiveness of the proposed algorithms on both synthetic and real-world datasets.
Source: arXiv (Cornell University)
Institutions: Pennsylvania State University, University of Virginia
Language: English
Field: Decision Sciences
Domain: Social Sciences
Open Access status: green
Awards: Collaborative Research: SWIFT: SMALL: Learning-Efficient Spectrum Access for No-Sensing Devices in Shared Spectrum, Collaborative Research: SWIFT: SMALL: Learning-Efficient Spectrum Access for No-Sensing Devices in Shared Spectrum, Collaborative Research: CNS Core: Small: Timely Computing and Learning over Communication Networks, Collaborative Research: MLWiNS: Dino-RL: A Domain Knowledge Enriched Reinforcement Learning Framework for Wireless Network Optimization, Collaborative Research: MLWiNS: Dino-RL: A Domain Knowledge Enriched Reinforcement Learning Framework for Wireless Network Optimization, CCSS: Collaborative Research: Towards a Resource Rationing Framework for Wireless Federated Learning, CNS Core: Medium: When Next Generation Wireless Networks Meet Machine Learning
OpenAlex ID: W3208202329