Federated Learning in Cooperative Healthcare with Resilient Aggregation for Dynamic Privacy Protection

Authors

  • Harini Subramaniam
  • Karan Kaur

DOI:

https://doi.org/10.71086/IAJSE/V11I3/IAJSE1154

Keywords:

Federated Learning, Privacy-Preserving, Machine Learning, Aggregation, Healthcare.

Abstract

In the last few years, the rising implementation of Federated Learning (FL) algorithms and increasing fears over
personal information privacy have garnered significant interest from researchers and businesses in Privacy-Preserving
(PP)-FL. Effective PP-FL enables several participants to independently develop their machine learning (ML) models,
which are then combined to form a universal framework while respecting privacy. This study introduced Federated
Learning in Cooperative Healthcare with Resilient Aggregation for Dynamic Privacy Protection (FL-RA-DPP). This
study implements and assesses several resilient aggregation techniques in FL utilized in medical information.
Moreover, it has been demonstrated that these approaches may identify and eliminate defective or malevolent local
clients throughout the learning process. Experiments have been conducted employing practical and
cooperative healthcare databases to learn medical diagnostic categorization tasks. The database has been employed to
model the efficacy of these distinct robust FL aggregation algorithms in response to various harming attempts. It has
been noted that DPP did not markedly influence the ultimate training convergence of the various aggregation
procedures.

Downloads

Published

2024-09-30

Issue

Section

Articles

How to Cite

Subramaniam, H., & Kaur, K. (2024). Federated Learning in Cooperative Healthcare with Resilient Aggregation for Dynamic Privacy Protection. International Academic Journal of Science and Engineering, 11(3), 15-18. https://doi.org/10.71086/IAJSE/V11I3/IAJSE1154