Graph Anomaly Detection on Computers (Anomaly-F1, Avg. Rank)
85.78Anomaly F1 ScoreN2NSC
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| N2NSCLLM Backbone=Qwen3-8B, PEFT=LoRA (rank 64), GNN Encoder=2-layer GAT, GNN heads=8, GNN output dimension=256, Contrastive selection K=22026.06 | 85.78 | 1 | |
| GINCategory=GNN-based methods2026.06 | 62.86 | 7 | |
| XGBGraphCategory=GNN-based methods2026.06 | 57.34 | 5 | |
| GCNCategory=GNN-based methods2026.06 | 57.19 | 7.25 | |
| RFGraphCategory=GNN-based methods2026.06 | 54.44 | 7.62 | |
| TAPECategory=Integrating LLMs with graphs2026.06 | 53.44 | 5.25 | |
| GATCategory=GNN-based methods2026.06 | 52.47 | 9.62 | |
| GraphSAGECategory=GNN-based methods2026.06 | 48.71 | 5.5 | |
| PC-GNNCategory=GNN-based methods2026.06 | 45.63 | 7.25 | |
| BWGNNCategory=GNN-based methods2026.06 | 34.71 | 8 | |
| GHRNCategory=GNN-based methods2026.06 | 33.66 | 9.25 | |
| InstructGLMCategory=Integrating LLMs with graphs2026.06 | 27.51 | 12 | |
| ConsisGADCategory=Integrating LLMs with graphs2026.06 | 26.94 | 14.12 | |
| GAAPCategory=GNN-based methods2026.06 | 25.5 | 12.25 | |
| PMPCategory=GNN-based methods2026.06 | 24.89 | 12.38 | |
| CoLLCategory=Integrating LLMs with graphs2026.06 | 22.91 | 14.62 | |
| UNPromptCategory=Integrating LLMs with graphs2026.06 | 14.3 | 15.12 | |
| GraphGPTCategory=Integrating LLMs with graphs2026.06 | 10.66 | 17.75 |