Summary of Stealing Training Graphs From Graph Neural Networks, by Minhua Lin et al.
Stealing Training Graphs from Graph Neural Networks
by Minhua Lin, Enyan Dai, Junjie Xu, Jinyuan Jia, Xiang Zhang, Suhang Wang
First submitted to arxiv on: 17 Nov 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: 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 A novel machine learning framework is presented, which enables the theft of training graphs from Graph Neural Networks (GNNs) with high accuracy. The framework utilizes a graph diffusion model to generate samples that resemble the target training set and leverages GNN model parameters to identify training graphs. Experimental results on real-world datasets demonstrate the effectiveness of this approach in stealing training graphs from trained GNNs. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A new way to steal information from special types of computer programs called Graph Neural Networks (GNNs) is discovered. GNNs are used for tasks like bioinformatics, but they can accidentally reveal private data. Researchers found that the parameters of these models are strongly connected to the training graphs, which means they can be used to steal this data. To achieve this, a new model called graph diffusion was developed, which generates fake graphs that resemble the real ones. The study shows how this framework can successfully steal training graphs from trained GNNs. |
Keywords
» Artificial intelligence » Diffusion » Diffusion model » Gnn » Machine learning
