Journal of Computer Applications ›› 0, Vol. ›› Issue (): 112-117.DOI: 10.11772/j.issn.1001-9081.2024030347

• Cyber security • Previous Articles     Next Articles

Decentralized federated learning privacy protection method for medical scenarios

Guilin GUAN1,2(), Zhengping TAO1,2, Ting ZHI1,2, Yang CAO1,2, Zhenqiang XIE1,2,3   

  1. 1.CETC Big Data Research Institute Company Limited,Guiyang Guizhou 550022,China
    2.National Engineering Research Center of Big Data Application on Improving Government Governance Capacities,Guiyang Guizhou 550022,China
    3.College of Intelligence and Computing,Tianjin University,Tianjin 300350,China
  • Received:2024-03-27 Revised:2024-05-20 Accepted:2024-05-27 Online:2025-01-24 Published:2024-12-31
  • Contact: Guilin GUAN

面向医疗场景的去中心化联邦学习隐私保护方法

管桂林1,2(), 陶政坪1,2, 支婷1,2, 曹扬1,2, 谢真强1,2,3   

  1. 1.中电科大数据研究院有限公司,贵阳 550022
    2.提升政府治理能力大数据应用技术国家工程研究中心,贵阳 550022
    3.天津大学 智能与计算学部,天津 300350
  • 通讯作者: 管桂林
  • 作者简介:管桂林(1994—),男,贵州普安人,工程师,硕士,主要研究方向:联邦学习、区块链、密码学、机器学习
    陶政坪(1995—),男,贵州天柱人,高级工程师,硕士,主要研究方向:隐私计算、联邦学习、人工智能
    支婷(1993—),女,贵州贵阳人,工程师,博士,主要研究方向:信息安全、自然语言处理、可信联邦学习
    曹扬(1981—),男,四川成都人,高级工程师,博士研究生,主要研究方向:信息安全、自然语言处理、可信联邦学习
    谢真强(1990—),男,贵州贵阳人,高级工程师,博士研究生,主要研究方向:隐私计算、网络安全、联邦学习。
  • 基金资助:
    国家重点研发计划项目(2023YFC3806001);贵州省科技支撑计划项目(2023MA6DN7B8X22057);贵州省科技重大专项(黔科合重大专项字[2024]002);2024年度中央引导地方科技发展资金资助项目(黔科合中引地[2024]009)

Abstract:

A decentralized federated learning privacy protection method for medical scenarios was proposed to address the issues of slow model training convergence, gradient information leakage, single point of failure and vulnerability to attacks of central servers in most existing federated learning schemes in medical industry, resulting in problem of low global model robustness. Firstly, an efficient federated learning method based on Fletcher-Reeves (FR) conjugate gradient algorithm was constructed to achieve fast convergence, secure and reliable computation during the model training process. And the Paillier homomorphic encryption algorithm was used to solve the problem of gradient data leakage in medical institutions. Secondly, a decentralized trusted federated learning architecture based on blockchain consensus mechanism was proposed, thereby achieving secure modeling without the coordination of a trusted central server, and avoiding the problem of low training efficiency of the central server caused by attacks or paralysis. At the same time, global and local model updates were stored in the blockchain to achieve full lifecycle protection of the model. Security analysis results indicate that the proposed method possesses indistinguishability of ciphertext and data privacy. Simulation results show the proposed method has the advantages of fast convergence and high accuracy compared to the comparison methods, and can meet the practical needs of the existing medical data joint modeling better.

Key words: federated learning, medical data, blockchain, decentralization, privacy protection

摘要:

针对医疗行业现有的联邦学习方案大多存在的由于模型训练收敛慢、梯度信息泄露、中央服务器的单点失效以及易受攻击而导致的全局模型鲁棒性低等问题,提出一种面向医疗场景的去中心化联邦学习隐私保护方法。首先,构建基于FR(Fletcher-Reeves)共轭梯度算法的高效联邦学习方法,从而实现模型训练过程中的快速收敛和安全可靠计算,并通过Paillier同态加密算法解决医疗机构梯度数据泄露的问题;其次,提出基于区块链共识机制的去中心化可信联邦学习架构,无须可信的中央服务器协调即可实现安全建模,并避免了由于中央服务器遭受攻击或者瘫痪导致的训练效率低下问题;同时,将全局模型和本地模型更新存储于区块链,从而实现对模型全生命周期的保护。安全性分析结果表明,所提方法具备密文不可区分性和数据隐私性。仿真实验结果表明,所提方法相较于对比方法具有收敛快、准确性高等优势,能较好地满足现有医疗数据联合建模的实际需求。

关键词: 联邦学习, 医疗数据, 区块链, 去中心化, 隐私保护

CLC Number: