Principal Neighbourhood Aggregation for Graph Nets

Abstract

Graph Neural Networks (GNNs) have been shown to be effective models for different predictive tasks on graph-structured data. Recent work on their expressive power has focused on isomorphism tasks and countable feature spaces. We extend this theoretical framework to include continuous features - which occur regularly in real-world input domains and within the hidden layers of GNNs - and we demonstrate the requirement for multiple aggregation functions in this context. Accordingly, we propose Principal Neighbourhood Aggregation (PNA), a novel architecture combining multiple aggregators with degree-scalers (which generalize the sum aggregator). Finally, we compare the capacity of different models to capture and exploit the graph structure via a novel benchmark containing multiple tasks taken from classical graph theory, alongside existing benchmarks from real-world domains, all of which demonstrate the strength of our model. With this work, we hope to steer some of the GNN research towards new aggregation methods which we believe are essential in the search for powerful and robust models.

34th Conference on Neural Information Processing Systems (NeurIPS 2020)

References in corpus (23)

• Statistical mechanics of complex networks
• Semi-Supervised Classification with Graph Convolutional Networks
• Inductive Representation Learning on Large Graphs
• Neural Message Passing for Quantum Chemistry
• Geometric deep learning: going beyond Euclidean data
• Spectral Networks and Locally Connected Networks on Graphs
• Open Graph Benchmark: Datasets for Machine Learning on Graphs
• Adaptive Computation Time for Recurrent Neural Networks
• Residual Gated Graph ConvNets
• Benchmarking Graph Neural Networks
• PairNorm: Tackling Oversmoothing in GNNs
• An Efficient Graph Convolutional Network Technique for the Travelling Salesman Problem
• Order Matters: Sequence to sequence for sets
• Adaptive Propagation Graph Convolutional Network
• A Survey on The Expressive Power of Graph Neural Networks
• Understanding the Representation Power of Graph Neural Networks in Learning Graph Topology
• Generalization and Representational Limits of Graph Neural Networks
• A simple yet effective baseline for non-attributed graph classification
• Relational Pooling for Graph Representations
• Can Graph Neural Networks Count Substructures?
• Improving Graph Attention Networks with Large Margin-based Constraints
• Directional Graph Networks
• Neural Execution of Graph Algorithms

Cited by in corpus (46)

• A Comprehensive Survey on Deep Graph Representation Learning
• Improving Graph Neural Network Expressivity via Subgraph Isomorphism Counting
• ETA Prediction with Graph Neural Networks in Google Maps
• Neural Bellman-Ford Networks: A General Graph Neural Network Framework for Link Prediction
• Global Self-Attention as a Replacement for Graph Convolution
• Deep learning of contagion dynamics on complex networks
• Rethinking Graph Transformers with Spectral Attention
• Anomaly Detection in Dynamic Graphs: A Comprehensive Survey
• KPGT: Knowledge-Guided Pre-training of Graph Transformer for Molecular Property Prediction
• High-Level Synthesis Performance Prediction using GNNs: Benchmarking, Modeling, and Advancing
• Learning the Travelling Salesperson Problem Requires Rethinking Generalization
• Graph Neural Networks for Graph Drawing
• Learning to Extend Molecular Scaffolds with Structural Motifs
• On the design space between molecular mechanics and machine learning force fields
• Designing the Topology of Graph Neural Networks: A Novel Feature Fusion Perspective
• Automated 3D Pre-Training for Molecular Property Prediction
• MM-GNN: Mix-Moment Graph Neural Network towards Modeling Neighborhood Feature Distribution
• Multi-task graph neural networks for simultaneous prediction of global and atomic properties in ferromagnetic systems
• Building powerful and equivariant graph neural networks with structural message-passing
• MoGERNN: An Inductive Traffic Predictor for Unobserved Locations
• How hard is to distinguish graphs with graph neural networks?
• Predicting cognitive scores with graph neural networks through sample selection learning
• GNNBuilder: An Automated Framework for Generic Graph Neural Network Accelerator Generation, Simulation, and Optimization
• Advancing Molecular Machine Learning Representations with Stereoelectronics-Infused Molecular Graphs
• A Graph Isomorphism Network with Weighted Multiple Aggregators for Speech Emotion Recognition
• Multiresolution Graph Transformers and Wavelet Positional Encoding for Learning Hierarchical Structures
• Parameterized Hypercomplex Graph Neural Networks for Graph Classification
• LineMVGNN: Anti-Money Laundering with Line-Graph-Assisted Multi-View Graph Neural Networks
• Improving the Expressiveness of $K$-hop Message-Passing GNNs by Injecting Contextualized Substructure Information
• DynamoRep: Trajectory-Based Population Dynamics for Classification of Black-box Optimization Problems
• Directional Graph Networks
• From Local Structures to Size Generalization in Graph Neural Networks
• Field-level simulation-based inference with galaxy catalogs: the impact of systematic effects
• Weisfeiler and Lehman Go Paths: Learning Topological Features via Path Complexes
• Graph neural induction of value iteration
• Co-attention Graph Pooling for Efficient Pairwise Graph Interaction Learning
• Neural Message Passing for Objective-Based Uncertainty Quantification and Optimal Experimental Design
• Complete Neural Networks for Complete Euclidean Graphs
• Breaking the Expressive Bottlenecks of Graph Neural Networks
• Adaptive Neural Message Passing for Inductive Learning on Hypergraphs
• PACIA: Parameter-Efficient Adapter for Few-Shot Molecular Property Prediction
• Message Passing in Graph Convolution Networks via Adaptive Filter Banks
• Learning Conjoint Attentions for Graph Neural Nets
• On Local Aggregation in Heterophilic Graphs
• LSP : Acceleration and Regularization of Graph Neural Networks via Locality Sensitive Pruning of Graphs
• Edge-featured Graph Neural Architecture Search