Federating for Learning Group Fair Models

TL;DR

This paper introduces FedMinMax, an algorithm for achieving group fairness in federated learning when participants have access to different demographic groups, with proven guarantees and experimental validation.

Contribution

It formalizes minmax group fairness in federated learning and proposes FedMinMax, an optimization algorithm with performance guarantees for this fairness criterion.

Findings

FedMinMax effectively improves group fairness across diverse federated setups.

The proposed method outperforms existing fairness approaches in experiments.

The algorithm maintains performance guarantees similar to centralized learning.

Abstract

Federated learning is an increasingly popular paradigm that enables a large number of entities to collaboratively learn better models. In this work, we study minmax group fairness in paradigms where different participating entities may only have access to a subset of the population groups during the training phase. We formally analyze how this fairness objective differs from existing federated learning fairness criteria that impose similar performance across participants instead of demographic groups. We provide an optimization algorithm -- FedMinMax -- for solving the proposed problem that provably enjoys the performance guarantees of centralized learning algorithms. We experimentally compare the proposed approach against other methods in terms of group fairness in various federated learning setups.