Federated Online Learning Based Recommendation Systems for Mobile Social Applications
Guanyi Su · 2021 IEEE Intl Conf on Dependable, Autonomic and Secure Computing, Intl Conf on Pervasive Intelligence and Computing, Intl Conf on Cloud and Big Data Computing, Intl Conf on Cyber Science and Technology Congress (DASC/PiCom/CBDCom/CyberSciTech) · 2021
With the explosion of big data, the conventional centralized service paradigm cannot meet the rapid demands of mobile social users. Social recommender systems emerge as an effective approach can boost the recommendation effectiveness by inferring user's interests and preferences. However, it is challenge to improve the accuracy of recommendation due to the diversity and complexity of big data. To improve the quality of experience (QoE) of user, in this paper, we propose a federated online learning scheme for recommendation systems to promote information sharing and improve the accuracy of recommendation. The system model for recommender systems in mobile social networks is first presented. Then, to encourage mobile clients to participant in local training, the contract-based incentive scheme is designed to increase the utilities of clients and the third party (TTP). Next, the federated online learning problem for recommendation systems in social community is then established. The federated online learning for recommender systems algorithm is proposed to derive the optimal model parameters. Finally, experiment results have been conducted to validate the efficiency of the proposed scheme.