Summary of Fast Inference Of Removal-based Node Influence, by Weikai Li et al.
Fast Inference of Removal-Based Node Influence
by Weikai Li, Zhiping Xiao, Xiao Luo, Yizhou Sun
First submitted to arxiv on: 13 Mar 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 method for evaluating node influence in graph neural networks (GNNs) is proposed, which measures the prediction change of a trained GNN model caused by removing a node. This approach has applications such as predicting Twitter accounts’ polarity if a particular account were removed. The authors use the GNN as a surrogate model to simulate the change of nodes or edges caused by node removal, and propose an efficient method called NORA (NOde-Removal-based fAst GNN inference) that uses gradient information to approximate the node-removal influence. NORA only requires one forward propagation and one backpropagation to estimate the influence score for all nodes. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Evaluating how much a social media account’s tweets affect other accounts’ opinions is a new challenge in using graph neural networks (GNNs). Researchers have been trying to solve this problem, but it was slow and not very efficient. A new method called NORA is introduced that makes it faster and easier to find out how much each node in the network affects others. This can be useful for predicting what would happen if a particular account were taken away. |
Keywords
* Artificial intelligence * Backpropagation * Gnn * Inference
