README.md

Publications in ICML 2022

Semi-supervised Node Classification

• TAM: Topology-Aware Margin Loss for Class-Imbalanced Node Classification
• Nonlinear Feature Diffusion on Hypergraphs

Graph Classification

• Structural Entropy Guided Graph Hierarchical Pooling
• A Hierarchical Transitive-Aligned Graph Kernel for Un-attributed Graphs
• The Infinite Contextual Graph Markov Model

Data Augmentation

• G-Mixup: Graph Data Augmentation for Graph Classification
• Learning from Counterfactual Links for Link Prediction
• Local Augmentation for Graph Neural Networks

Scalability

• GraphFM: Improving Large-Scale GNN Training via Feature Momentum
• Generalization Guarantee of Training Graph Convolutional Networks with Graph Topology Sampling
• Scalable Deep Gaussian Markov Random Fields for General Graphs
• LeNSE: Learning To Navigate Subgraph Embeddings for Large-Scale Combinatorial Optimisation

Explainability

• Efficient Computation of Higher-Order Subgraph Attribution via Message Passing
• Information Bottleneck-Guided Stochastic Attention Mechanism for Interpretable Graph Learning

Oversmoothing/Depth

• Optimization-induced Implicit Graph Diffusion
• pathGCN: Learning General Graph Spatial Operators from Paths
• NAFS: A Simple yet Tough-to-beat Baseline for Graph Representation Learning
• Graph-Coupled Oscillator Networks

Heterophily

• Finding Global Homophily in Graph Neural Networks When Meeting Heterophily
• $p$-Laplacian Based Graph Neural Networks
• G$^2$CN: Graph Gaussian Convolution Networks with Concentrated Graph Filters

Spectral Methods

• Convergence of Invariant Graph Networks
• A New Perspective on the Effects of Spectrum in Graph Neural Networks
• How Powerful are Spectral Graph Neural Networks

Expressivity

• SPECTRE : Spectral Conditioning Overcomes the Expressivity Limits of One-shot Graph Generators
• A Theoretical Comparison of Graph Neural Network Extensions
• Boosting Graph Structure Learning with Dummy Nodes
• Weisfeiler-Lehman meets Gromov-Wasserstein
• Going Deeper into Permutation-Sensitive Graph Neural Networks

Temporal Data

• DSTAGNN: Dynamic Spatial-Temporal Aware Graph Neural Network for Traffic Flow Forecasting
• On the Equivalence Between Temporal and Static Equivariant Graph Representations

Equivariance

• Equivariant graph neural networks with complete local frames
• Equivariant Quantum Graph Circuits
• SpeqNets: Sparsity-aware permutation-equivariant graph networks

Molecules

• 3DLinker: An E(3) Equivariant Variational Autoencoder for Molecular Linker Design
• 3D Infomax improves GNNs for Molecular Property Prediction
• Molecular Graph Representation Learning via Heterogeneous Motif Graph Construction
• Generative Coarse-Graining of Molecular Conformations
• Equivariant Diffusion for Molecule Generation in 3D

Proteins

• Generating 3D Molecules for Target Protein Binding
• Pocket2Mol: Efficient Molecular Sampling Based on 3D Protein Pockets
• Antibody-Antigen Interface Design via Hierarchical Structure Refinement

Generative Models

• VarScene: A Deep Generative Model for Realistic Scene Graph Synthesis
• Score-based Generative Modeling of Graphs via the System of Stochastic Differential Equations

Transformers

• Structure-Aware Transformer for Graph Representation Learning
• Retroformer: Pushing the Limits of End-to-end Retrosynthesis Transformer
• From block-Toeplitz matrices to differential equations on graphs: towards a general theory for scalable masked Transformers

Knowledge Bases

• Neural-Symbolic Models for Logical Queries on Knowledge Graphs
• Cycle Representation Learning for Inductive Relation Prediction
• Semiparametric Subgraph Reasoning for Question Answering over Large Knowledge Bases

Self-Supervision

• Omni-Granular Ego-Semantic Propagation for Self-Supervised Graph Representation Learning
• Self-Supervised Representation Learning via Latent Graph Prediction

Contrastive Learning

• ProGCL: Rethinking Hard Negative Mining in Graph Contrastive Learning
• Augment with Care: Contrastive Learning for Combinatorial Problems
• Let Invariant Rationale Discovery Inspire Graph Contrastive Learning

Active Learning

• Cross-Space Active Learning on Graph Convolutional Networks
• GALAXY: Graph-based Active Learning at the Extreme

Reinforcement Learning

• Topology-Aware Network Pruning using Multi-stage Graph Embedding and Reinforcement Learning
• Optimizing Tensor Network Contraction Using Reinforcement Learning
• Model-based Meta Reinforcement Learning using Graph Structured Surrogate Models and Amortized Policy Search
• Learning to Assemble with Large-Scale Structured Reinforcement Learning

Physics

• Constraint-based graph network simulator
• Learning to Solve PDE-constrained Inverse Problems with Graph Networks

Architecture Search

• Deep and Flexible Graph Neural Architecture Search
• Large-Scale Graph Neural Architecture Search

Anomaly Detection

• Rethinking Graph Neural Networks for Anomaly Detection
• Deep Variational Graph Convolutional Recurrent Network for Multivariate Time Series Anomaly Detection

Directed Graphs

• PACE: A Parallelizable Computation Encoder for Directed Acyclic Graphs
• GNNRank: Learning Global Rankings from Pairwise Comparisons via Directed Graph Neural Networks

Benchmarks

• The CLRS Algorithmic Reasoning Benchmark
• Particle Transformer for Jet Tagging

Miscellaneous

• GACT: Activation Compressed Training for Generic Network Architectures
• Unsupervised Detection of Contextualized Embedding Bias with Application to Ideology
• Injecting Logical Constraints into Neural Networks via Straight-Through Estimators
• Learning Iterative Reasoning through Energy Minimization