FedML-HE: An Efficient Homomorphic-Encryption-Based Privacy-Preserving Federated Learning System

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Jin, Weizhao, Yao, Yuhang, Han, Shanshan, Gu, Jiajun, Joe-Wong, Carlee, Ravi, Srivatsan, Avestimehr, Salman, He, Chaoyang
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917695560089600
author Jin, Weizhao
Yao, Yuhang
Han, Shanshan
Gu, Jiajun
Joe-Wong, Carlee
Ravi, Srivatsan
Avestimehr, Salman
He, Chaoyang
author_facet Jin, Weizhao
Yao, Yuhang
Han, Shanshan
Gu, Jiajun
Joe-Wong, Carlee
Ravi, Srivatsan
Avestimehr, Salman
He, Chaoyang
contents Federated Learning trains machine learning models on distributed devices by aggregating local model updates instead of local data. However, privacy concerns arise as the aggregated local models on the server may reveal sensitive personal information by inversion attacks. Privacy-preserving methods, such as homomorphic encryption (HE), then become necessary for FL training. Despite HE's privacy advantages, its applications suffer from impractical overheads, especially for foundation models. In this paper, we present FedML-HE, the first practical federated learning system with efficient HE-based secure model aggregation. FedML-HE proposes to selectively encrypt sensitive parameters, significantly reducing both computation and communication overheads during training while providing customizable privacy preservation. Our optimized system demonstrates considerable overhead reduction, particularly for large foundation models (e.g., ~10x reduction for ResNet-50, and up to ~40x reduction for BERT), demonstrating the potential for scalable HE-based FL deployment.
format Preprint
id arxiv_https___arxiv_org_abs_2303_10837
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle FedML-HE: An Efficient Homomorphic-Encryption-Based Privacy-Preserving Federated Learning System
Jin, Weizhao
Yao, Yuhang
Han, Shanshan
Gu, Jiajun
Joe-Wong, Carlee
Ravi, Srivatsan
Avestimehr, Salman
He, Chaoyang
Machine Learning
Cryptography and Security
Federated Learning trains machine learning models on distributed devices by aggregating local model updates instead of local data. However, privacy concerns arise as the aggregated local models on the server may reveal sensitive personal information by inversion attacks. Privacy-preserving methods, such as homomorphic encryption (HE), then become necessary for FL training. Despite HE's privacy advantages, its applications suffer from impractical overheads, especially for foundation models. In this paper, we present FedML-HE, the first practical federated learning system with efficient HE-based secure model aggregation. FedML-HE proposes to selectively encrypt sensitive parameters, significantly reducing both computation and communication overheads during training while providing customizable privacy preservation. Our optimized system demonstrates considerable overhead reduction, particularly for large foundation models (e.g., ~10x reduction for ResNet-50, and up to ~40x reduction for BERT), demonstrating the potential for scalable HE-based FL deployment.
title FedML-HE: An Efficient Homomorphic-Encryption-Based Privacy-Preserving Federated Learning System
topic Machine Learning
Cryptography and Security
url https://arxiv.org/abs/2303.10837