Towards fair and personalized federated recommendation

Shanfeng Wang, Hao Tao, Jianzhao Li, Xinyuan Ji, Yuan Gao, Maoguo Gong

Abstract: Highlights•We design a novel federated recommendation framework, which takes fairness and personalization of recommendation into consideration simultaneously.•In the proposed framework, user local model, cluster-level model, and global model are adopted to obtain personalized recommendation.•On the client, FFFR adopts graph neural network and filter network to better learn users and items representations.