Summary of Explaining Graph Neural Networks For Node Similarity on Graphs, by Daniel Daza et al.
Explaining Graph Neural Networks for Node Similarity on Graphs
by Daniel Daza, Cuong Xuan Chu, Trung-Kien Tran, Daria Stepanova, Michael Cochez, Paul Groth
First submitted to arxiv on: 10 Jul 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 A novel study explores explainable similarity search over graphs, focusing on augmenting Graph Neural Network (GNN)-based methods for computing node similarities with explanations. The research evaluates two prominent approaches: mutual information (MI) and gradient-based explanations (GB). Results show that GB explanations possess three desirable properties: actionability, consistency, and pruning ability, making them a promising solution for similarity search applications in citation networks or knowledge graphs. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Similarity search is important for finding patterns in graph data. Graph Neural Networks (GNNs) are good at this, but they don’t always tell us why two things are similar. This paper looks at how to add explanations to GNN-based methods. The researchers test two ways: mutual information and gradient-based explanations. They find that the second method is better because it helps make predictions, works consistently, and can be simplified without losing its power. |
Keywords
* Artificial intelligence * Gnn * Graph neural network * Pruning
