Summary of Deep Efficient Private Neighbor Generation For Subgraph Federated Learning, by Ke Zhang et al.
Deep Efficient Private Neighbor Generation for Subgraph Federated Learning
by Ke Zhang, Lichao Sun, Bolin Ding, Siu Ming Yiu, Carl Yang
First submitted to arxiv on: 9 Jan 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
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 the challenge of federated learning (FL) for graph mining models in realistic applications where data is fragmented across multiple owners. The subgraph FL scenario involves local clients holding subgraphs of the global graph, with the goal of obtaining globally generalized models without compromising data privacy. The authors propose FedDEP, a novel approach that addresses three key challenges: incomplete information propagation on local subgraphs due to missing cross-subgraph neighbors, utility efficiency, and privacy goals. FedDEP consists of deep neighbor generation through GNN embeddings, efficient pseudo-FL for neighbor generation, and noise-less edge-local-differential-privacy protection. Theoretical guarantees are provided for the correctness and efficiency of FedDEP, with empirical results on four real-world datasets demonstrating its benefits. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about a way to learn from big graphs that are split up into smaller pieces by different people. It’s like trying to figure out how all the puzzle pieces fit together without seeing the whole picture. The authors came up with a new method called FedDEP that helps solve this problem. They made three main changes: they created fake neighbor nodes based on what they know about the missing connections, they used a special kind of learning called pseudo-FL to make it more efficient, and they added extra protection to keep the data safe. The authors tested their method on four real-world datasets and showed that it works well. |
Keywords
* Artificial intelligence * Federated learning * Gnn
