└── README.md /README.md: -------------------------------------------------------------------------------- 1 | # Graph OOD Detection 2 | 3 | A curated list of resources for OOD detection with graph data. 4 | If some related papers are missing, please contact us via pull requests :) 5 | 6 | ### Relation to the anomaly detection problem 7 | 8 | Unlike traditional anomaly detection, which assumes a small percentage of anomalies, 9 | in the graph domain, the OOD part may contain nodes that are comparable in size to the ID part. 10 | 11 | (taken from "Learning on Graphs with Out-of-Distribution Nodes", KDD 2022) 12 | 13 | Anomaly graphs are usually malicious samples from a real system and available during training; while OOD graphs are samples from the distribution that the model has not seen during training. 14 | 15 | (taken from "A Data-centric Framework to Endow Graph Neural Networks with Out-Of-Distribution Detection Ability", KDD 2023) 16 | 17 | 18 | ## Recent works 19 | ### Graph-level OOD detection 20 | KDD 2023. A Data-centric Framework to Endow Graph Neural Networks with Out-Of-Distribution Detection Ability. 21 | [[paper]](http://shichuan.org/doc/150.pdf) 22 | [[code]](https://github.com/BUPT-GAMMA/AAGOD) 23 | >- Graph-level, post-hoc, training without outlier exposure. 24 | 25 | WSDM 2023. GOOD-D: On Unsupervised Graph Out-Of-Distribution Detection 26 | [[paper]](https://arxiv.org/pdf/2211.04208.pdf) 27 | [[code]](https://github.com/yixinliu233/G-OOD-D) 28 | >- Graph-level, Self-supervised contrastive learning. 29 | 30 | 🔥 NeurIPS 2022. GraphDE: A Generative Framework for Debiased Learning and Out-of-Distribution Detection on Graphs. 31 | [[paper]](https://openreview.net/pdf?id=mSiPuHIP7t8) 32 | [[code]](https://github.com/Emiyalzn/GraphDE) 33 | >- Graph-level, Generative, Variational inference. 34 | 35 | ICML workshop 2022. Towards OOD Detection in Graph Classification from Uncertainty Estimation Perspective. 36 | [[paper]](https://arxiv.org/pdf/2206.10691.pdf) 37 | >- Graph-level, Uncertainty estimation. 38 | 39 | ### Node-level OOD detection 40 | 🔥 ICLR 2023. Energy-based out-of-distribution detection for graph neural networks. 41 | [[paper]](https://arxiv.org/pdf/2302.02914.pdf) 42 | [[code]](https://github.com/qitianwu/GraphOOD-GNNSafe) 43 | >- Node-level with two disconnected graphs, Energy-based score with pagerank-based propagation. 44 | 45 | KDD 2022. Learning on Graphs with Out-of-Distribution Nodes. 46 | [[paper]](https://dl.acm.org/doi/10.1145/3534678.3539457) 47 | [[code]](https://github.com/SongYYYY/KDD22-OODGAT) 48 | >- Node-level with inter-edges, Semi-supervised outlier detection, a modified GAT with three regularization terms. 49 | 50 | NeurIPS 2021. Graph Posterior Network: Bayesian Predictive Uncertainty for Node Classification. 51 | [[paper]](https://proceedings.neurips.cc/paper/2021/file/95b431e51fc53692913da5263c214162-Paper.pdf) 52 | >- Node-level, Bayesian posterior. 53 | 54 | NeurIPS 2020. Uncertainty Aware Semi-Supervised Learning on Graph Data. 55 | [[paper]](https://proceedings.neurips.cc/paper/2020/file/968c9b4f09cbb7d7925f38aea3484111-Paper.pdf) 56 | [[code]](https://github.com/zxj32/uncertainty-GNN) 57 | >- Node-level, a Graph-based Kernel Dirichlet distribution Estimation (GKDE) method. 58 | 59 | ### Anomaly Detection 60 | NeurIPS 2023. GADBench: Revisiting and Benchmarking Supervised Graph Anomaly Detection. 61 | [[paper]](https://arxiv.org/pdf/2306.12251.pdf) 62 | [[code]](https://github.com/squareRoot3/GADBench) 63 | 64 | ICLR 2023 submission. Revisiting Uncertainty Estimation for Node Classification: New Benchmark and Insights 65 | [[paper]](https://openreview.net/pdf?id=DB3BH3arU2Y) 66 | 67 | NeurIPS 2022. Dual-discriminative Graph Neural Network for Imbalanced Graph-level Anomaly Detection. 68 | [[paper]](https://openreview.net/pdf?id=d6mf9AFoR-O) 69 | [[code]](https://github.com/graph-level-anomalies/iGAD) 70 | 71 | 🔥 ICML 2022. Rethinking Graph Neural Networks for Anomaly Detection. 72 | [[paper]](https://proceedings.mlr.press/v162/tang22b/tang22b.pdf) 73 | [[code]](https://github.com/squareRoot3/Rethinking-Anomaly-Detection) 74 | >- Anomalies leads to the ‘right-shift’ of spectral energy. 75 | 76 | IJCAI 2022. Raising the Bar in Graph-level Anomaly Detection. 77 | [[paper]](https://arxiv.org/pdf/2205.13845.pdf) 78 | [[code]](https://github.com/boschresearch/GraphLevel-AnomalyDetection) 79 | 80 | WSDM 2022. Deep Graph-level Anomaly Detection by Glocal Knowledge Distillation. 81 | [[paper]](https://arxiv.org/pdf/2112.10063.pdf) 82 | [[code]](https://github.com/RongrongMa/GLocalKD) 83 | >- Graph-level Anomaly Detection, Joint random distillation of graph and node representations with two GNNs (GLocalKD). 84 | 85 | Big Data 2021. On Using Classification Datasets to Evaluate Graph Outlier Detection: Peculiar Observations and New Insights. 86 | [[paper]](https://arxiv.org/pdf/2012.12931.pdf) 87 | >- Graph-level, Propagation-based outlier detection method (OCGIN). 88 | 89 | [pygod-team/pygod](https://github.com/pygod-team/pygod) 90 | 91 | ### OOD Generalization 92 | - Arxiv 2022. Out-Of-Distribution Generalization on Graphs: A Survey. 93 | [paper](https://arxiv.org/pdf/2202.07987.pdf) 94 | - [THUMNLab/awesome-graph-ood](https://github.com/THUMNLab/awesome-graph-ood) 95 | - [Out-Of-Distribution Generalization on Graphs: Paper List](https://graph.ood-generalization.com/) 96 | --------------------------------------------------------------------------------