Federated Learning with Non-IID Data

TLDR

Federated learning lets resource‑constrained edge devices train a shared model locally, offering privacy, security, regulatory and economic benefits. The study addresses the statistical challenge of federated learning with non‑IID local data. The authors propose creating a small globally shared data subset to improve training on non‑IID data. Experiments reveal that non‑IID data can reduce federated learning accuracy by up to 55%, a drop explained by weight divergence measured via earth‑mover distance, but adding just 5% globally shared data raises accuracy by 30% on CIFAR‑10.

Abstract

Federated learning enables resource-constrained edge compute devices, such as mobile phones and IoT devices, to learn a shared model for prediction, while keeping the training data local. This decentralized approach to train models provides privacy, security, regulatory and economic benefits. In this work, we focus on the statistical challenge of federated learning when local data is non-IID. We first show that the accuracy of federated learning reduces significantly, by up to 55% for neural networks trained for highly skewed non-IID data, where each client device trains only on a single class of data. We further show that this accuracy reduction can be explained by the weight divergence, which can be quantified by the earth mover's distance (EMD) between the distribution over classes on each device and the population distribution. As a solution, we propose a strategy to improve training on non-IID data by creating a small subset of data which is globally shared between all the edge devices. Experiments show that accuracy can be increased by 30% for the CIFAR-10 dataset with only 5% globally shared data.