Summary of Understanding the Impact Of Graph Reduction on Adversarial Robustness in Graph Neural Networks, by Kerui Wu et al.
Understanding the Impact of Graph Reduction on Adversarial Robustness in Graph Neural Networks
by Kerui Wu, Ka-Ho Chow, Wenqi Wei, Lei Yu
First submitted to arxiv on: 8 Dec 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 study explores the susceptibility of Graph Neural Networks (GNNs) to adversarial attacks when using graph reduction techniques for scalability. The investigation examines the impact of graph coarsening and sparsification on the robustness of GNNs against poisoning attacks, using multiple datasets and architectures. Results show that while sparsification can mitigate some attacks, it has limited effect on others; coarsening tends to amplify adversarial impact, reducing classification accuracy with decreasing reduction ratio. The study also analyzes causes driving these effects and evaluates defensive GNN models under graph reduction. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary GNNs are special computers designed for big data problems. This paper looks at how well they work when trying to fight fake information attacks on large networks. Researchers tested many ways to make the GNNs smaller, like taking out some parts or making connections between nodes weaker. They found that sometimes this helps and sometimes it makes things worse. The study also explains why these methods can be good or bad and what we can do to make our GNNs stronger. |
Keywords
* Artificial intelligence * Classification * Gnn
