Differentially Private Client Selection and Resource Allocation in Federated Learning for Medical Applications Using Graph Neural Networks

被引:0
|
作者
Messinis, Sotirios C. [1 ]
Protonotarios, Nicholas E. [2 ]
Doulamis, Nikolaos [1 ]
机构
[1] Natl Tech Univ Athens, Inst Commun & Comp Syst, Athens 15773, Greece
[2] Acad Athens, Math Res Ctr, Athens 11527, Greece
关键词
decentralized federated learning; resource allocation; differential privacy; client selection; graph neural networks; AGGREGATION;
D O I
10.3390/s24165142
中图分类号
O65 [分析化学];
学科分类号
070302 ; 081704 ;
摘要
Federated learning (FL) has emerged as a pivotal paradigm for training machine learning models across decentralized devices while maintaining data privacy. In the healthcare domain, FL enables collaborative training among diverse medical devices and institutions, enhancing model robustness and generalizability without compromising patient privacy. In this paper, we propose DPS-GAT, a novel approach integrating graph attention networks (GATs) with differentially private client selection and resource allocation strategies in FL. Our methodology addresses the challenges of data heterogeneity and limited communication resources inherent in medical applications. By employing graph neural networks (GNNs), we effectively capture the relational structures among clients, optimizing the selection process and ensuring efficient resource distribution. Differential privacy mechanisms are incorporated, to safeguard sensitive information throughout the training process. Our extensive experiments, based on the Regensburg pediatric appendicitis open dataset, demonstrated the superiority of our approach, in terms of model accuracy, privacy preservation, and resource efficiency, compared to traditional FL methods. The ability of DPS-GAT to maintain a high and stable number of client selections across various rounds and differential privacy budgets has significant practical implications, indicating that FL systems can achieve strong privacy guarantees without compromising client engagement and model performance. This balance is essential for real-world applications where both privacy and performance are paramount. This study suggests a promising direction for more secure and efficient FL medical applications, which could improve patient care through enhanced predictive models and collaborative data utilization.
引用
收藏
页数:18
相关论文
共 50 条
  • [31] Investigating the Predictive Reproducibility of Federated Graph Neural Networks Using Medical Datasets
    Balik, Mehmet Yigit
    Rekik, Arwa
    Rekik, Islem
    PREDICTIVE INTELLIGENCE IN MEDICINE (PRIME 2022), 2022, 13564 : 160 - 171
  • [32] PriFairFed: A Local Differentially Private Federated Learning Algorithm for Client-Level Fairness
    Hu, Chuang
    Wu, Nanxi
    Shi, Siping
    Liu, Xuan
    Luo, Bing
    Wang, Kanye Ye
    Jiang, Jiawei
    Cheng, Dazhao
    IEEE TRANSACTIONS ON MOBILE COMPUTING, 2025, 24 (05) : 3993 - 4005
  • [33] Federated Learning under Channel Uncertainty: Joint Client Scheduling and Resource Allocation
    Wadu, Madhusanka Manimel
    Samarakoon, Sumudu
    Bennis, Mehdi
    2020 IEEE WIRELESS COMMUNICATIONS AND NETWORKING CONFERENCE (WCNC), 2020,
  • [34] Blockchain-Aided Wireless Federated Learning: Resource Allocation and Client Scheduling
    Li, Jun
    Zhang, Weiwei
    Wei, Kang
    Chen, Guangji
    Shu, Feng
    Chen, Wen
    Jin, Shi
    IEEE INTERNET OF THINGS JOURNAL, 2024, 11 (21): : 34349 - 34363
  • [35] Joint Client Scheduling and Resource Allocation Under Channel Uncertainty in Federated Learning
    Wadu, Madhusanka Manimel
    Samarakoon, Sumudu
    Bennis, Mehdi
    IEEE TRANSACTIONS ON COMMUNICATIONS, 2021, 69 (09) : 5962 - 5974
  • [36] Joint Client Selection and Bandwidth Allocation of Wireless Federated Learning by Deep Reinforcement Learning
    Mao, Wei
    Lu, Xingjian
    Jiang, Yuhui
    Zheng, Haikun
    IEEE TRANSACTIONS ON SERVICES COMPUTING, 2024, 17 (01) : 336 - 348
  • [37] GAP: Differentially Private Graph Neural Networks with Aggregation Perturbation
    Sajadmanesh, Sina
    Shamsabadi, Ali Shahin
    Bellet, Aurelien
    Gatica-Perez, Daniel
    PROCEEDINGS OF THE 32ND USENIX SECURITY SYMPOSIUM, 2023, : 3223 - 3240
  • [38] GAP: Differentially Private Graph Neural Networks with Aggregation Perturbation
    Sajadmanesh, Sina
    Shamsabadi, Ali Shahin
    Bellet, Aurélien
    Gatica-Perez, Daniel
    arXiv, 2022,
  • [39] A Federated Learning Approach for Graph Convolutional Neural Networks
    Campbell, Andrew
    Liu, Hang
    Scaglione, Anna
    Wu, Tong
    2024 IEEE 13RD SENSOR ARRAY AND MULTICHANNEL SIGNAL PROCESSING WORKSHOP, SAM 2024, 2024,
  • [40] Decentralized Wireless Resource Allocation with Graph Neural Networks
    Wang, Zhiyang
    Eisen, Mark
    Ribeiro, Alejandro
    2020 54TH ASILOMAR CONFERENCE ON SIGNALS, SYSTEMS, AND COMPUTERS, 2020, : 299 - 303