Size-Invariant Graph Representations for Graph Classification Extrapolations
arXiv:2103.05045
Abstract
In general, graph representation learning methods assume that the train and test data come from the same distribution. In this work we consider an underexplored area of an otherwise rapidly developing field of graph representation learning: The task of out-of-distribution (OOD) graph classification, where train and test data have different distributions, with test data unavailable during training. Our work shows it is possible to use a causal model to learn approximately invariant representations that better extrapolate between train and test data. Finally, we conclude with synthetic and real-world dataset experiments showcasing the benefits of representations that are invariant to train/test distribution shifts.
arXiv commentAccepted at ICML 2021
References in corpus (48)
- Scikit-learn: Machine Learning in Python
- PyTorch: An Imperative Style, High-Performance Deep Learning Library
- Inductive Representation Learning on Large Graphs
- Neural Message Passing for Quantum Chemistry
- Shortcut Learning in Deep Neural Networks
- Higher-order organization of complex networks
- Fast Graph Representation Learning with PyTorch Geometric
- Representation Learning on Graphs with Jumping Knowledge Networks
- Variational Graph Auto-Encoders
- Biological network comparison using graphlet degree distribution
- Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representations
- Interaction Networks for Learning about Objects, Relations and Physics
- Underspecification Presents Challenges for Credibility in Modern Machine Learning
- A Survey on Graph Kernels
- TUDataset: A collection of benchmark datasets for learning with graphs
- Invariant Risk Minimization
- Improving Graph Neural Network Expressivity via Subgraph Isomorphism Counting
- WILDS: A Benchmark of in-the-Wild Distribution Shifts
- Causal Effect Inference with Deep Latent-Variable Models
- Neural Combinatorial Optimization with Reinforcement Learning
- Hyperbolic Graph Convolutional Neural Networks
- Out-of-Distribution Generalization via Risk Extrapolation (REx)
- Mapping multiplex hubs in human functional brain network
- Measuring abstract reasoning in neural networks
- A Meta-Transfer Objective for Learning to Disentangle Causal Mechanisms
- Causality for Machine Learning
- Stochastic blockmodel approximation of a graphon: Theory and consistent estimation
- Invariant and Equivariant Graph Networks
- Data augmentation instead of explicit regularization
- Machine Learning on Graphs: A Model and Comprehensive Taxonomy
- The Risks of Invariant Risk Minimization
- Learning Independent Causal Mechanisms
- A Survey on The Expressive Power of Graph Neural Networks
- Hyperbolic Graph Neural Networks
- Causal Discovery from Changes
- Learning the Travelling Salesperson Problem Requires Rethinking Generalization
- Higher-order Graph Convolutional Networks
- Rogue Signs: Deceiving Traffic Sign Recognition with Malicious Ads and Logos
- Telling cause from effect in deterministic linear dynamical systems
- Causal Generative Neural Networks
- GraphCrop: Subgraph Cropping for Graph Classification
- Group invariance principles for causal generative models
- Estimating Generalization under Distribution Shifts via Domain-Invariant Representations
- On the Equivalence Between Temporal and Static Graph Representations for Observational Predictions
- Heterogeneous Network Motifs
- From Local Structures to Size Generalization in Graph Neural Networks
- Representation Learning of Graphs Using Graph Convolutional Multilayer Networks Based on Motifs
- Neural Networks for Learning Counterfactual G-Invariances from Single Environments