A Federated Learning Framework for Healthcare IoT devices

TL;DR

This paper proposes a federated learning framework tailored for healthcare IoT devices that reduces communication overhead and maintains high accuracy despite limited device resources.

Contribution

It introduces a partitioned neural network training approach with activation and gradient sparsification to optimize federated learning for resource-constrained healthcare IoT devices.

Findings

Achieves 0.2% of the synchronization traffic of vanilla federated learning.

Maintains low accuracy loss with reduced communication overhead.

Empirical results validate the efficiency of the proposed framework.

Abstract

The Internet of Things (IoT) revolution has shown potential to give rise to many medical applications with access to large volumes of healthcare data collected by IoT devices. However, the increasing demand for healthcare data privacy and security makes each IoT device an isolated island of data. Further, the limited computation and communication capacity of wearable healthcare devices restrict the application of vanilla federated learning. To this end, we propose an advanced federated learning framework to train deep neural networks, where the network is partitioned and allocated to IoT devices and a centralized server. Then most of the training computation is handled by the powerful server. The sparsification of activations and gradients significantly reduces the communication overhead. Empirical study have suggested that the proposed framework guarantees a low accuracy loss, while only requiring 0.2% of the synchronization traffic in vanilla federated learning.