Summary of Generation Is Better Than Modification: Combating High Class Homophily Variance in Graph Anomaly Detection, by Rui Zhang et al.
Generation is better than Modification: Combating High Class Homophily Variance in Graph Anomaly Detection
by Rui Zhang, Dawei Cheng, Xin Liu, Jie Yang, Yi Ouyang, Xian Wu, Yefeng Zheng
First submitted to arxiv on: 15 Mar 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 a new metric called Class Homophily Variance to quantify differences in homophily distribution between classes in graph anomaly detection. To mitigate this impact, it proposes a novel GNN model named HedGe that generates new relationships with low class homophily variance. This approach uses self-attention mechanisms and leverages nodes relevant in the feature space but not directly connected in the original graph. The paper also modifies the loss function to punish unnecessary heterophilic edges. Extensive comparison experiments show HedGe achieves best performance across multiple benchmark datasets, including anomaly detection and edgeless node classification. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Graphs can be tricky! This paper looks at how to spot weird things (anomalies) in graphs. They found that when we try to find these anomalies, the way things are connected changes a lot between different groups of points. To deal with this, they made a new tool called HedGe that creates new connections to help find anomalies better. It works by looking at how similar points are in the graph and making new connections that are helpful for finding weird things. This paper shows that HedGe does a great job on lots of different datasets. |
Keywords
* Artificial intelligence * Anomaly detection * Classification * Gnn * Loss function * Self attention
