Summary of Gisexplainer: on Explainability Of Graph Neural Networks Via Game-theoretic Interaction Subgraphs, by Xingping Xian et al.
GISExplainer: On Explainability of Graph Neural Networks via Game-theoretic Interaction Subgraphs
by Xingping Xian, Jianlu Liu, Chao Wang, Tao Wu, Shaojie Qiao, Xiaochuan Tang, Qun Liu
First submitted to arxiv on: 24 Sep 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Social and Information Networks (cs.SI)
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 proposed GISExplainer method offers a game-theoretic interaction-based explanation technique to uncover what Graph Neural Networks (GNNs) have learned for node classification by discovering human-interpretable causal explanatory subgraphs. This approach considers the causal interaction between edges within different coalition scales, providing faithful explanations. The method defines a causal attribution mechanism that quantifies the effect of an edge on predictions and assumes that coalitions with negative effects are significant for model interpretation. A sequential decision process selects salient edges to form an explanatory subgraph, aiming for better explanations. An efficiency optimization scheme is also proposed through coalition sampling. GISExplainer outperforms state-of-the-art approaches in terms of Fidelity and Sparsity. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary GISExplainer is a new way to understand how Graph Neural Networks (GNNs) work. It’s like trying to figure out why a doctor made a certain diagnosis or an investor chose a particular stock. The method helps by showing how individual pieces of information interact with each other to create the GNN’s prediction. This can be very important in fields like healthcare, finance, and cybersecurity where decisions need to be transparent. GISExplainer is better than existing methods at giving faithful explanations. |
Keywords
* Artificial intelligence * Classification * Gnn * Optimization
