Summary of Revisiting Early-learning Regularization When Federated Learning Meets Noisy Labels, by Taehyeon Kim et al.
Revisiting Early-Learning Regularization When Federated Learning Meets Noisy Labels
by Taehyeon Kim, Donggyu Kim, Se-Young Yun
First submitted to arxiv on: 8 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Distributed, Parallel, and Cluster Computing (cs.DC)
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 tackles a crucial challenge in federated learning (FL): addressing label noise when data is collected across multiple clients. Traditional approaches to mitigate label noise are limited by privacy concerns and the diversity of client data. The authors revisit early-learning regularization, introducing Federated Label-mixture Regularization (FLR). FLR generates new pseudo labels by combining local and global model predictions. This method improves the accuracy of the global model in both identical and non-identical datasets, while also countering memorization of noisy labels. FLR is compatible with existing label noise and FL techniques, paving the way for better generalization in FL environments plagued by inaccurate labels. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary In this paper, scientists try to solve a big problem in artificial intelligence called federated learning. When many devices collect data together, it’s hard to make sure all that information is accurate. The authors come up with a new idea called Federated Label-mixture Regularization (FLR). FLR helps by combining what each device thinks the correct answer is and what the global model thinks it is. This makes the global model more accurate and helps it ignore noisy or incorrect labels. This breakthrough could lead to better AI models that work well in real-world situations. |
Keywords
* Artificial intelligence * Federated learning * Generalization * Regularization
