Summary of Cluster-guided Contrastive Class-imbalanced Graph Classification, by Wei Ju et al.
Cluster-guided Contrastive Class-imbalanced Graph Classification
by Wei Ju, Zhengyang Mao, Siyu Yi, Yifang Qin, Yiyang Gu, Zhiping Xiao, Jianhao Shen, Ziyue Qiao, Ming Zhang
First submitted to arxiv on: 17 Dec 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); 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 A novel approach called C^3GNN is proposed for class-imbalanced graph classification, which effectively addresses the limitations of existing methods in vision and graph neural networks (GNNs). The issue with GNNs is that they tend to predict classes biased towards the majority, while existing class-imbalanced learning methods may overlook semantic substructures in majority classes. C^3GNN integrates clustering into contrastive learning to enhance classification. It clusters graphs from each majority class into multiple subclasses, similar in size to minority classes, mitigating imbalance. Mixup technique generates synthetic samples to enrich subclass diversity. Supervised contrastive learning learns hierarchical graph representations, exploring semantic substructures while avoiding excessive focus on minority classes. Experiments on real-world datasets demonstrate the superior performance of C^3GNN over competitive baselines. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary C^3GNN is a new way to classify graphs when some categories have much more data than others. This problem makes it hard for computer programs to learn from all the data equally well. Some methods try to solve this by focusing on the minority classes, but they might miss important details in the majority classes. C^3GNN solves this by grouping together similar graphs in the majority categories and creating fake examples to help train the model. This helps the computer program learn more about all types of graphs. |
Keywords
» Artificial intelligence » Classification » Clustering » Gnn » Supervised
