GPT-FL: Generative Pre-trained Model-Assisted Federated Learning

TL;DR

GPT-FL introduces a federated learning framework that utilizes generative pre-trained models to produce synthetic data, enhancing model accuracy, communication efficiency, and convergence speed across various data modalities.

Contribution

This work presents a novel GPT-FL framework that leverages generative pre-trained models to improve federated learning performance with synthetic data training.

Findings

GPT-FL outperforms state-of-the-art FL methods in accuracy and efficiency.

Synthetic data generated by GPT-FL enhances convergence speed.

GPT-FL achieves significant gains regardless of data domain alignment.

Abstract

In this work, we propose GPT-FL, a generative pre-trained model-assisted federated learning (FL) framework. At its core, GPT-FL leverages generative pre-trained models to generate diversified synthetic data. These generated data are used to train a downstream model on the server, which is then fine-tuned with private client data under the standard FL framework. We show that GPT-FL consistently outperforms state-of-the-art FL methods in terms of model test accuracy, communication efficiency, and client sampling efficiency. Through comprehensive ablation analysis across various data modalities, we discover that the downstream model generated by synthetic data plays a crucial role in controlling the direction of gradient diversity during FL training, which enhances convergence speed and contributes to the notable accuracy boost observed with GPT-FL. Also, regardless of whether the target data falls within or outside the domain of the pre-trained generative model, GPT-FL consistently achieves significant performance gains, surpassing the results obtained by models trained solely with FL or synthetic data. The code is available at https://github.com/AvestimehrResearchGroup/GPT-FL.