Generative Risk Minimization for Out-of-Distribution Generalization on Graphs
Not on the map. Why?
Abstract (arXiv)
Out-of-distribution (OOD) generalization on graphs aims at dealing with scenarios where the test graph distribution differs from the training graph distributions. Compared to i.i.d. data like images, the OOD generalization problem on graph-structured data remains challenging due to the non-i.i.d. property and complex structural information on graphs. Recently, several works on graph OOD generalization have explored extracting invariant subgraphs that share crucial classification information across different distributions. Nevertheless, such a strategy could be suboptimal for entirely capturing the invariant information, as the extraction of discrete structures could potentially lead to the loss of invariant information or the involvement of spurious information. In this paper, we propose an innovative framework, named Generative Risk Minimization (GRM), designed to generate an invariant subgraph for each input graph to be classified, instead of extraction. To address the challenge of optimization in the absence of optimal invariant subgraphs (i.e., ground truths), we derive a tractable form of the proposed GRM objective by introducing a latent causal variable, and its effectiveness is validated by our theoretical analysis. We further conduct extensive experiments across a variety of real-world graph datasets for both node-level and graph-level OOD generalization, and the results demonstrate the superiority of our framework GRM.
Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 58 of its 72 references that are in the map. First posted on arXiv 11 Feb 2025.
Builds on
58 of 72 references.
References · 58
- 2016Semi-Supervised Classification with Graph Convolutional Networkscited by 11,282
- 2018Representation Learning with Contrastive Predictive Codingcited by 7,662
- 2017Inductive Representation Learning on Large Graphscited by 6,284
- 2017mixup: Beyond Empirical Risk Minimizationcited by 5,155
- 2018How Powerful are Graph Neural Networks?cited by 4,355
- 2020Supervised Contrastive Learningcited by 2,803
- 2020Open Graph Benchmark: Datasets for Machine Learning on Graphscited by 2,148
- 2014Unsupervised Domain Adaptation by Backpropagationcited by 1,906
- 2019Invariant Risk Minimizationcited by 1,832
- 2018Graph Neural Networks: A Review of Methods and Applicationscited by 1,571
- 2019Do ImageNet Classifiers Generalize to ImageNet?cited by 1,507
- 2016Variational Graph Auto-Encoderscited by 1,373
38 more
- 2020WILDS: A Benchmark of in-the-Wild Distribution Shiftscited by 1,366
- 2018Moment Matching for Multi-Source Domain Adaptationcited by 1,327
- 2017MoleculeNet: A Benchmark for Molecular Machine Learningcited by 1,037
- 2019An Introduction to Variational Autoencoderscited by 997
- 2020Out-of-Distribution Generalization via Risk Extrapolation (REx)cited by 737
- 2019GNNExplainer: Generating Explanations for Graph Neural Networkscited by 735
- 2018Recognition in Terra Incognitacited by 655
- 2018Junction Tree Variational Autoencoder for Molecular Graph Generationcited by 647
- 2013Domain Generalization via Invariant Feature Representationcited by 622
- 2017One pixel attack for fooling deep neural networkscited by 579
- 2018GraphRNN: Generating Realistic Graphs with Deep Auto-regressive Modelscited by 406
- 2018MolGAN: An implicit generative model for small molecular graphscited by 392
- 2015Return of Frustratingly Easy Domain Adaptationcited by 389
- 2020Explainability in Graph Neural Networks: A Taxonomic Surveycited by 383
- 2018GraphVAE: Towards Generation of Small Graphs Using Variational Autoencoderscited by 375
- 2020Environment Inference for Invariant Learningcited by 306
- 2018Fr'echet ChemNet Distance: A metric for generative models for molecules in drug discoverycited by 191
- 2020Invariant Risk Minimization Gamescited by 188
- 2022Discovering Invariant Rationales for Graph Neural Networkscited by 186
- 2020Adaptive Risk Minimization: Learning to Adapt to Domain Shiftcited by 183
- 2020Design Space for Graph Neural Networkscited by 182
- 2022Interpretable and Generalizable Graph Learning via Stochastic Attention Mechanismcited by 171
- 2019Understanding Attention and Generalization in Graph Neural Networkscited by 169
- 2022Handling Distribution Shifts on Graphs: An Invariance Perspectivecited by 163
- 2022Learning Causally Invariant Representations for Out-of-Distribution Generalization on Graphscited by 144
- 2022GOOD: A Graph Out-of-Distribution Benchmarkcited by 140
- 2020Invariant Rationalizationcited by 131
- 2018Dark Model Adaptation: Semantic Image Segmentation from Daytime to Nighttimecited by 122
- 2018Graphite: Iterative Generative Modeling of Graphscited by 112
- 2022Out-Of-Distribution Generalization on Graphs: A Surveycited by 97
- 2020Graph Meta Learning via Local Subgraphscited by 89
- 2021Size-Invariant Graph Representations for Graph Classification Extrapolationscited by 82
- 2023Virtual Node Tuning for Few-shot Node Classificationcited by 46
- 2020Transductive Information Maximization For Few-Shot Learningcited by 44
- 2023Does Invariant Graph Learning via Environment Augmentation Learn Invariance?cited by 42
- 2021Graph Few-shot Class-incremental Learningcited by 40
- 2023Mind the Label Shift of Augmentation-based Graph OOD Generalizationcited by 34
- 2022Transductive Linear Probing: A Novel Framework for Few-Shot Node Classificationcited by 24
50 most cited of 58 All 58 (JSON)
Built on by
Citing papers per year: 2026: 2.
2 papers.