Position-aware Graph Neural Networks

Abstract

Learning node embeddings that capture a node's position within the broader graph structure is crucial for many prediction tasks on graphs. However, existing Graph Neural Network (GNN) architectures have limited power in capturing the position/location of a given node with respect to all other nodes of the graph. Here we propose Position-aware Graph Neural Networks (P-GNNs), a new class of GNNs for computing position-aware node embeddings. P-GNN first samples sets of anchor nodes, computes the distance of a given target node to each anchor-set,and then learns a non-linear distance-weighted aggregation scheme over the anchor-sets. This way P-GNNs can capture positions/locations of nodes with respect to the anchor nodes. P-GNNs have several advantages: they are inductive, scalable,and can incorporate node feature information. We apply P-GNNs to multiple prediction tasks including link prediction and community detection. We show that P-GNNs consistently outperform state of the art GNNs, with up to 66% improvement in terms of the ROC AUC score.

ICML 2019, long oral

Cited by in corpus (52)

• MAGNN: Metapath Aggregated Graph Neural Network for Heterogeneous Graph Embedding
• Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gauges
• Open Graph Benchmark: Datasets for Machine Learning on Graphs
• A Generalization of Transformer Networks to Graphs
• Benchmarking Graph Neural Networks
• Strategies for Pre-training Graph Neural Networks
• Design Space for Graph Neural Networks
• Graph Neural Networks: Taxonomy, Advances and Trends
• Do Transformers Really Perform Bad for Graph Representation?
• Distance Encoding: Design Provably More Powerful Neural Networks for Graph Representation Learning
• Graph Information Bottleneck
• Handling Missing Data with Graph Representation Learning
• GraphSAIL: Graph Structure Aware Incremental Learning for Recommender Systems
• Residual Correlation in Graph Neural Network Regression
• Accurate Learning of Graph Representations with Graph Multiset Pooling
• Federated Graph Classification over Non-IID Graphs
• Generalization and Representational Limits of Graph Neural Networks
• Graph Clustering with Graph Neural Networks
• Topology-Imbalance Learning for Semi-Supervised Node Classification
• Subgraph Neural Networks
• Graph Neural Networks for Graph Drawing
• Hashing-Accelerated Graph Neural Networks for Link Prediction
• Heterogeneous Network Representation Learning: A Unified Framework with Survey and Benchmark
• Graph Neural Networks with Learnable Structural and Positional Representations
• Meta-Learning with Graph Neural Networks: Methods and Applications
• Multiresolution Graph Transformers and Wavelet Positional Encoding for Learning Hierarchical Structures
• On the Equivalence of Decoupled Graph Convolution Network and Label Propagation
• On Positional and Structural Node Features for Graph Neural Networks on Non-attributed Graphs
• On Graph Neural Networks versus Graph-Augmented MLPs
• Hierarchical Message-Passing Graph Neural Networks
• Utilising Graph Machine Learning within Drug Discovery and Development
• Knowledge-aware Coupled Graph Neural Network for Social Recommendation
• Missing Data Estimation in Temporal Multilayer Position-aware Graph Neural Network (TMP-GNN)
• Permutohedral-GCN: Graph Convolutional Networks with Global Attention
• Reconstruction for Powerful Graph Representations
• Parameter Prediction for Unseen Deep Architectures
• Learnable Structural Semantic Readout for Graph Classification
• Graph Inference Representation: Learning Graph Positional Embeddings with Anchor Path Encoding
• AdaGNN: A multi-modal latent representation meta-learner for GNNs based on AdaBoosting
• Cross-Node Federated Graph Neural Network for Spatio-Temporal Data Modeling
• Graph Attention Networks with Positional Embeddings
• Edgeless-GNN: Unsupervised Representation Learning for Edgeless Nodes
• Robust Hierarchical Graph Classification with Subgraph Attention
• Unsupervised Graph Representation by Periphery and Hierarchical Information Maximization
• Learning Parametrised Graph Shift Operators
• Graph Embedding VAE: A Permutation Invariant Model of Graph Structure
• BiTe-GCN: A New GCN Architecture via BidirectionalConvolution of Topology and Features on Text-Rich Networks
• Isometric Graph Neural Networks
• Heterogeneous Grid Convolution for Adaptive, Efficient, and Controllable Computation
• A Survey on Role-Oriented Network Embedding
• ROPE: Reading Order Equivariant Positional Encoding for Graph-based Document Information Extraction
• HyperExpan: Taxonomy Expansion with Hyperbolic Representation Learning