Hierarchically Fair Federated Learning
share
Federated learning facilitates collaboration among self-interested agents. However, agents participate only if they are fairly rewarded. To encourage participation, this paper introduces a new fairness notion based on the proportionality principle, i.e., more contribution should lead to more reward. To achieve this, we propose a novel hierarchically fair federated learning (HFFL) framework. Under this framework, agents are rewarded in proportion to their contributions which properly incentivizes collaboration. HFFL+ extends this to incorporate heterogeneous models. Theoretical analysis and empirical evaluation on several datasets confirm the efficacy of our frameworks in upholding fairness.
READ FULL TEXT
