Summary of Gradient Inversion Attack on Graph Neural Networks, by Divya Anand Sinha et al.
Gradient Inversion Attack on Graph Neural Networks
by Divya Anand Sinha, Yezi Liu, Ruijie Du, Yanning Shen
First submitted to arxiv on: 29 Nov 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 In this paper, researchers investigate the vulnerability of graph data and graph neural networks (GNNs) to attacks during federated learning. They study whether private data can be recovered from leaked gradients in both node classification and graph classification tasks, proposing a novel attack named Graph Leakage from Gradients (GLG). The authors analyze two widely-used GNN frameworks, GCN and GraphSAGE, discussing the effects of different model settings on recovery. Through theoretical analysis and empirical validation, they show that parts of the graph data can be leaked from the gradients. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about keeping private information safe when using artificial intelligence to learn patterns in big graphs. Imagine many people working together to train a computer program without sharing their individual data. This is called federated learning, and it’s important for protecting privacy. However, researchers found that some bad actors could steal private data by looking at the changes made during this process. The authors of this paper looked at how this might happen with graph neural networks (GNNs), which are special kinds of artificial intelligence designed to work with graphs. They tested two popular GNN models and showed that parts of the graph data can be leaked. |
Keywords
» Artificial intelligence » Classification » Federated learning » Gcn » Gnn
