Federated machine learning for edge computing

Abstract

Federated Learning (FL) has emerged as a promising approach for training machine learning models across distributed nodes without requiring raw data centralization. This research explores the feasibility of deploying federated machine learning models on microcontrollers. The research investigates the deployment of lightweight C++ clients on resource-constrained devices and the integration of a Python-based central server with these clients. The proposed solution contributes to the advancement of edge intelligence by enabling efficient on-device model training and inference.