Summary of Fedmlp: Federated Multi-label Medical Image Classification Under Task Heterogeneity, by Zhaobin Sun (1) et al.
FedMLP: Federated Multi-Label Medical Image Classification under Task Heterogeneity
by Zhaobin Sun, Nannan Wu, Junjie Shi, Li Yu, Xin Yang, Kwang-Ting Cheng, Zengqiang Yan
First submitted to arxiv on: 27 Jun 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary In this paper, researchers propose a novel approach to address the challenges of multi-label medical image classification in decentralized organizations with varying levels of medical knowledge and disease prevalence. The authors introduce a realistic label missing setting in cross-silo federated learning (FL) and develop a two-stage method, FedMLP, which consists of pseudo label tagging and global knowledge learning. FedMLP leverages warmed-up models to generate class prototypes and select samples with high confidence for missing labels, while utilizing a global model as a teacher for consistency regularization. The authors demonstrate the superiority of FedMLP against state-of-the-art approaches in federated semi-supervised and noisy label learning under task heterogeneity using publicly-available medical datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about how doctors and hospitals can work together to diagnose diseases from medical images without sharing their patient data. They want to do this by training a special kind of computer model that can understand different types of medical images, but each hospital may not have all the information they need to make accurate diagnoses. The researchers developed a new method called FedMLP to help solve this problem. It’s like a game where each hospital uses their own data and knowledge to train a smaller part of the big computer model, which helps everyone work together better. |
Keywords
» Artificial intelligence » Federated learning » Image classification » Regularization » Semi supervised
