Summary of Towards Federated Graph Learning in One-shot Communication, by Guochen Yan et al.
Towards Federated Graph Learning in One-shot Communication
by Guochen Yan, Xunkai Li, Luyuan Xie, Wentao Zhang, Qingni Shen, Yuejian Fang, Zhonghai Wu
First submitted to arxiv on: 18 Nov 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 Graph Learning (FGL) approach called One-shot Personalized Federated Graph Learning (O-pFGL), designed to train personalized models for node classification on distributed private graphs. O-pFGL addresses challenges in existing methods, such as communication overhead and bias, by estimating and aggregating class-wise feature distribution statistics to construct a global pseudo-graph. This method combines Secure Aggregation protocols for privacy preservation and a two-stage personalized training approach to balance local personal information and global insights from the pseudo-graph. Experimental results on 12 multi-scale graph datasets show that O-pFGL outperforms state-of-the-art baselines across various settings. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine you have many private graphs, like social networks or online communities, and you want to train a model that can work well on all of them without sharing the data. This is called Federated Graph Learning (FGL). But existing methods are not good at handling different types of graphs or require too much communication. The authors propose a new method called One-shot Personalized Federated Graph Learning (O-pFGL) that trains personalized models for node classification on these private graphs. It does this by creating a global model based on the features of all the graphs, which helps to balance the differences between them and improves the performance. |
Keywords
* Artificial intelligence * Classification * One shot
