代码地址:GitHub - yixinliu233/SIGNET: [NeurIPS'23] Towards Self-Interpretable Graph-Level Anomaly Detection 摘要 本研究旨在解决图级异常检测(Graph-level Anomaly Detection, GLAD)问题,即在图集合中识别出与大多数图显著不同的异常图。现有的工作主要集中在评估图级异常,但未能提供有意义的预测解释,这限制了...
本文的框架通过联合蒸馏每个图中 graph-level 和 node-level 的知识来学习全局和局部的图正则化模式信息,因此其包含了两个 GNN 和 两个损失函数,其一为固定的随机初始化权重的目标 GNN,另一个训练得到的预测 GNN,两者的网络结构相同。 GLocalKD 通过训练 predictor GNN h 来预测 random GNN 产生的图(节点)表示...
Abou Rida A, Amhaz R, Parrend P (2022) Evaluation of anomaly detection for cybersecurity using inductive node embedding with convolutional graph neural networks. Complex networks & their applications X: volume 2, proceedings of the tenth international conference on complex networks and their applicat...
Finally, a high-quality graph-level representation is learned by SubGAD for graph-level anomaly detection. Extensive experiments on 13 real-world datasets demonstrate the superiority of SubGAD compared withthe state-of-the-art methods. Our codes are availableat https://github.com/scu-kdde/OAM-...
(Molaei, Havvaei, Zare, & Jalili, 2021), node clustering (Molaei, Bousejin, Zare, & Jalili, 2021) (Molaei, Ghanbari Bousejin, et al. 2021), anomalous citation detection (Liu, Xia, Feng, Ren & Liu, 2022), epidemic prediction (Jin et al., 2023), and anomaly detection (Ren, Xia...
Even though this paper is focused on presenting a novel graph-based methodology to address graph-related class imbalance to improve node behaviour classification using supervised approaches, it is interesting to compare our results with traditional anomaly detection approaches that can be directly applied...
Paper tables with annotated results for Motif-Consistent Counterfactuals with Adversarial Refinement for Graph-Level Anomaly Detection
Graph anomaly detectiongraph neural networksgraph transfer learningExisting graph level anomaly detection methods are predominantly unsupervised due to high costs for obtaining labels, yielding sub-optimal detection accuracy when compared to supervised methods. Moreover, they heavily rely on the assumption ...
Graph level anomaly detection (GLAD) aims to spot anomalous graphs that structure pattern and feature information are different from most normal graphs in a graph set, which is rarely studied by other researchers but has significant application value. For instance, GLAD can be used to distinguish ...
ENTROPYGraph anomaly detection is crucial in many high-impact applications across diverse fields. In anomaly detection tasks, collecting plenty of annotated data tends to be costly and laborious. As a result, few-shot learning has been explored to address the issue by requiring only a fe...