Federated Model Aggregation via Self-Supervised Priors for Highly Imbalanced Medical Image Classification

Marawan Elbatel, Hualiang Wang, Robert Martí, Huazhu Fu, Xiaomeng Li
7/27/2023
cs.CVcs.LG

Official code: xmed-lab/Fed-MAS (17 stars, last commit 2023-12-02).

Abstract

In the medical field, federated learning commonly deals with highly imbalanced datasets, including skin lesions and gastrointestinal images. Existing federated methods under highly imbalanced datasets primarily focus on optimizing a global model without incorporating the intra-class variations that can arise in medical imaging due to different populations, findings, and scanners. In this paper, we study the inter-client intra-class variations with publicly available self-supervised auxiliary networks. Specifically, we find that employing a shared auxiliary pre-trained model, like MoCo-V2, locally on every client yields consistent divergence measurements. Based on these findings, we derive a dynamic balanced model aggregation via self-supervised priors (MAS) to guide the global model optimization. Fed-MAS can be utilized with different local learning methods for effective model aggregation toward a highly robust and unbiased global model. Our code is available at \url{https://github.com/xmed-lab/Fed-MAS}.

View on arXivView PDF

Code Implementations(1)

xmed-lab/Fed-MASOfficial100%

MICCAI 2023 DeCaF Best Paper Award: Federated Model Aggregation via Self-Supervised Priors for Highly Imbalanced Medical Image Classification

171PythonJul 25, 20232 years ago

Similar Papers

FedSoup: Improving Generalization and Personalization in Federated Learning via Selective Model Interpolation

85% match1 repo182023

Federated Learning for Medical Image Classification: A Comprehensive Benchmark

85% match1 repo12025

Class-Specific Distribution Alignment for Semi-Supervised Medical Image Classification

84% match2023

FedBN: Federated Learning on Non-IID Features via Local Batch Normalization

84% match2 repos7.2k2021

FedBM: Stealing Knowledge from Pre-trained Language Models for Heterogeneous Federated Learning

84% match1 repo22025

Global Intervention and Distillation for Federated Out-of-Distribution Generalization

84% match2025

Cite this paper

@article{elbatel2023federated,
  title  = {Federated Model Aggregation via Self-Supervised Priors for Highly Imbalanced Medical Image Classification},
  author = {Marawan Elbatel and Hualiang Wang and Robert Martí and Huazhu Fu and Xiaomeng Li},
  year   = {2023},
  eprint = {2307.14959},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2307.14959v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2307.14959.svg)](https://codeofpaper.com/paper/2307.14959)

Discussion