Summary of Eig-search: Generating Edge-induced Subgraphs For Gnn Explanation in Linear Time, by Shengyao Lu et al.
EiG-Search: Generating Edge-Induced Subgraphs for GNN Explanation in Linear Time
by Shengyao Lu, Bang Liu, Keith G. Mills, Jiao He, Di Niu
First submitted to arxiv on: 2 May 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 The paper introduces EiG-Search, a training-free approach for explaining Graph Neural Networks (GNNs) that induces subgraph explanations by edges. The goal is to balance intuitiveness and efficiency while ensuring transparency, addressing challenges faced by existing explainers that induce subgraphs by nodes. The approach employs an efficient linear-time search algorithm over edge-induced subgraphs ranked by an enhanced gradient-based importance. Experiments on seven datasets demonstrate superior performance and efficiency over leading baselines. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary GNNs are used to make predictions, but it’s hard to understand why they’re making those predictions. One way to explain this is to look at smaller parts of the graph that the GNN is using. But finding these parts can be tricky because it involves searching through a lot of data. This paper introduces a new way to do this search that is faster and more accurate than previous methods. It works by looking at edges between nodes in the graph, rather than just individual nodes. The approach also considers the importance of each edge in determining what’s important for explanation. The results show that this method outperforms other approaches on a range of datasets. |
Keywords
» Artificial intelligence » Gnn
