Summary of Tackling Data Heterogeneity in Federated Learning Via Loss Decomposition, by Shuang Zeng et al.
Tackling Data Heterogeneity in Federated Learning via Loss Decomposition
by Shuang Zeng, Pengxin Guo, Shuai Wang, Jianbo Wang, Yuyin Zhou, Liangqiong Qu
First submitted to arxiv on: 22 Aug 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 This paper proposes a novel Federated Learning (FL) method, FedLD, to mitigate the impact of data heterogeneity among clients in collaborative machine learning. The authors decompose the global loss into three terms: local loss, distribution shift loss, and aggregation loss. They find that existing FL methods focus on reducing either the distribution shift loss or the aggregation loss, but neglect the comprehensive joint effort to minimize all three terms. To address this gap, FedLD involves margin control regularization in local training to reduce the distribution shift loss and principal gradient-based server aggregation strategy to reduce the aggregation loss. The authors demonstrate that FedLD achieves better and more robust performance on retinal and chest X-ray classification compared to other FL algorithms. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Federated Learning (FL) helps doctors share medical data without sharing personal information. But when different hospitals have different patient records, it can be hard to get a good result. Researchers studied how FL works and found that most methods only try to fix one problem at a time. They created a new method called FedLD that tackles all three problems together. This helps FL work better, especially when dealing with differences in patient data. The researchers tested their method on pictures of retinas and chest X-rays and found it was more accurate and reliable than other methods. |
Keywords
» Artificial intelligence » Classification » Federated learning » Machine learning » Regularization
