Summary of Towards Cross-domain Few-shot Graph Anomaly Detection, by Jiazhen Chen et al.
Towards Cross-domain Few-shot Graph Anomaly Detection
by Jiazhen Chen, Sichao Fu, Zhibin Zhang, Zheng Ma, Mingbin Feng, Tony S. Wirjanto, Qinmu Peng
First submitted to arxiv on: 11 Oct 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 This paper proposes a novel approach to few-shot graph anomaly detection (GAD) in cross-domain scenarios, where the goal is to identify anomalies within sparsely labeled target graphs using auxiliary graphs from a related domain. The proposed framework, called CDFS-GAD, consists of three main components: a domain-adaptive graph contrastive learning module for enhancing cross-domain feature alignment, a prompt tuning module for extracting domain-specific features, and a domain-adaptive hypersphere classification loss for discrimination between normal and anomalous instances under minimal supervision. The self-training strategy is also employed to refine the predicted scores, improving its reliability in few-shot settings. Experimental results on twelve real-world datasets demonstrate the effectiveness of CDFS-GAD compared to existing GAD methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research paper focuses on a new way to detect unusual patterns in big data networks. Normally, we need lots of labeled training data to make accurate predictions. But what if we only have a few examples? This is the challenge that scientists are trying to solve. They’re developing a special method called CDFS-GAD that can work even when the test data is very different from the training data. The approach has three parts: first, it helps align features across domains; second, it adjusts the model to fit each domain’s specific needs; and third, it uses a special loss function to make accurate predictions with minimal supervision. The results show that this method works better than other existing methods on real-world datasets. |
Keywords
» Artificial intelligence » Alignment » Anomaly detection » Classification » Few shot » Loss function » Prompt » Self training
