Vertical Federated Linear Contextual Bandits

TL;DR

This paper introduces a novel vertical federated learning approach for contextual bandits, using a new encryption scheme to preserve privacy while maintaining high recommendation accuracy and efficiency.

Contribution

The paper proposes the orthogonal matrix-based mask mechanism (O3M) for privacy-preserving encryption in vertical federated bandits, enabling effective decentralized online recommendations.

Findings

The protocols recover centralized bandit performance with high accuracy.

The method offers strong privacy protection and efficiency.

Experimental results demonstrate superior recommendation quality.

Abstract

In this paper, we investigate a novel problem of building contextual bandits in the vertical federated setting, i.e., contextual information is vertically distributed over different departments. This problem remains largely unexplored in the research community. To this end, we carefully design a customized encryption scheme named orthogonal matrix-based mask mechanism(O3M) for encrypting local contextual information while avoiding expensive conventional cryptographic techniques. We further apply the mechanism to two commonly-used bandit algorithms, LinUCB and LinTS, and instantiate two practical protocols for online recommendation under the vertical federated setting. The proposed protocols can perfectly recover the service quality of centralized bandit algorithms while achieving a satisfactory runtime efficiency, which is theoretically proved and analyzed in this paper. By conducting extensive experiments on both synthetic and real-world datasets, we show the superiority of the proposed method in terms of privacy protection and recommendation performance.