Rethinking Contrastive Learning in Graph Anomaly Detection: A Clean-View Perspective

Authors: Di Jin, Jingyi Cao, Xiaobao Wang, Bingdao Feng, Dongxiao He, Longbiao Wang, Jianwu Dang

IJCAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments on five benchmark datasets validate the effectiveness of our approach. 4 Experiments 4.1 Experimental Setup Datasets. We conduct experiments on five benchmark datasets: Cora, Citeseer, Pub Med [Sen et al., 2008], Citation and ACM [Tang et al., 2008]. 4.3 Ablation Study To verify the effectiveness of different modules, we conduct three types of ablation studies.
Researcher Affiliation Academia 1College of Intelligence and Computing, Tianjin University, Tianjin, China 2Key Laboratory of Artificial Intelligence Application Technology, Qinghai Minzu University, China 3Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China
Pseudocode No The paper describes the methodology using prose and mathematical equations but does not include a distinct pseudocode or algorithm block.
Open Source Code No The paper does not provide any statements regarding the availability of open-source code, nor does it include links to code repositories.
Open Datasets Yes Datasets. We conduct experiments on five benchmark datasets: Cora, Citeseer, Pub Med [Sen et al., 2008], Citation and ACM [Tang et al., 2008].
Dataset Splits No The paper describes how anomalies are injected into the datasets but does not provide specific details on training, validation, and test splits for the experimental data.
Hardware Specification No The paper does not provide specific hardware details such as GPU models, CPU types, or memory used for running the experiments.
Software Dependencies No The paper does not provide specific software dependency details with version numbers, such as programming languages, libraries, or frameworks used.
Experiment Setup Yes Parameter Settings. In CVGAD, we employ a one-layer GCN to aggregate information from subgraphs, with both subgraph embeddings and node embeddings mapped to 64-dimensional vectors. The size of the subgraph is set to 4, and the learning rate remains fixed at 0.001. Additionally, we set the value of γ to 0.8. Five iterations are performed on all datasets. Specifically, for Cora, Cite Seer, and Pub Med, we perform edge removal every 100 epochs, conducting a total of 500 epochs. For Citation and ACM, we perform 1000 epochs of edge removal. In the refine training phase, we conduct 200 epochs on Cora, Cite Seer, and Pub Med, and 400 epochs on Citation and ACM. Besides, we implement 300 rounds of score calculation in all datasets. In addition, we set K to 0.015 for ACM and 0.01 for other datasets.