IP-FL: Incentivized and Personalized Federated Learning

  • 2024-10-11 18:44:56
  • Ahmad Faraz Khan, Xinran Wang, Qi Le, Zain ul Abdeen, Azal Ahmad Khan, Haider Ali, Ming Jin, Jie Ding, Ali R. Butt, Ali Anwar
  • 0

Abstract

Existing incentive solutions for traditional Federated Learning (FL) focus onindividual contributions to a single global objective, neglecting the nuancesof clustered personalization with multiple cluster-level models and thenon-monetary incentives such as personalized model appeal for clients. In thispaper, we first propose to treat incentivization and personalization asinterrelated challenges and solve them with an incentive mechanism that fosterspersonalized learning. Additionally, current methods depend on an aggregatorfor client clustering, which is limited by a lack of access to clients'confidential information due to privacy constraints, leading to inaccurateclustering. To overcome this, we propose direct client involvement, allowingclients to indicate their cluster membership preferences based on datadistribution and incentive-driven feedback. Our approach enhances thepersonalized model appeal for self-aware clients with high-quality data leadingto their active and consistent participation. Our evaluation demonstratessignificant improvements in test accuracy (8-45%), personalized model appeal(3-38%), and participation rates (31-100%) over existing FL models, includingthose addressing data heterogeneity and personalization.

 

Quick Read (beta)

loading the full paper ...