Summary of Rethinking the Representation in Federated Unsupervised Learning with Non-iid Data, by Xinting Liao et al.
Rethinking the Representation in Federated Unsupervised Learning with Non-IID Data
by Xinting Liao, Weiming Liu, Chaochao Chen, Pengyang Zhou, Fengyuan Yu, Huabin Zhu, Binhui Yao, Tao Wang, Xiaolin Zheng, Yanchao Tan
First submitted to arxiv on: 25 Mar 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 The proposed Federated Unsupervised Learning (FUSL) method, called FedU2, addresses the issue of insufficient representations in FUSL with non-IID data. The existing methods suffer from representation collapse entanglement among local and global models, as well as inconsistent representation spaces among local models. FedU2 consists of a flexible uniform regularizer (FUR) and an efficient unified aggregator (EUA). FUR disperses samples uniformly in each client to avoid representation collapse, while EUA constrains consistent client model updating on the server side. The performance of FedU2 is extensively validated through cross-device and cross-silo evaluation experiments on CIFAR10 and CIFAR100 datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary FedU2 is a new way for computers to learn from lots of different data sources without needing labels. This makes it harder for them to understand what they’re seeing, but FedU2 helps by making sure each source’s understanding is similar to the others. The method has two parts: one that keeps each source’s information separate and another that makes sure all the information is working together. Scientists tested this method on some famous data sets and found it worked really well. |
Keywords
* Artificial intelligence * Unsupervised
