Diffusion Improves Graph Learning

Abstract

Graph convolution is the core of most Graph Neural Networks (GNNs) and usually approximated by message passing between direct (one-hop) neighbors. In this work, we remove the restriction of using only the direct neighbors by introducing a powerful, yet spatially localized graph convolution: Graph diffusion convolution (GDC). GDC leverages generalized graph diffusion, examples of which are the heat kernel and personalized PageRank. It alleviates the problem of noisy and often arbitrarily defined edges in real graphs. We show that GDC is closely related to spectral-based models and thus combines the strengths of both spatial (message passing) and spectral methods. We demonstrate that replacing message passing with graph diffusion convolution consistently leads to significant performance improvements across a wide range of models on both supervised and unsupervised tasks and a variety of datasets. Furthermore, GDC is not limited to GNNs but can trivially be combined with any graph-based model or algorithm (e.g. spectral clustering) without requiring any changes to the latter or affecting its computational complexity. Our implementation is available online.

Published as a conference paper at NeurIPS 2019. Author name changed from Johannes Klicpera to Johannes Gasteiger

Cited by in corpus (24)

• Graph Contrastive Learning with Adaptive Augmentation
• Graph Self-Supervised Learning: A Survey
• Anomaly Detection on Attributed Networks via Contrastive Self-Supervised Learning
• A Comprehensive Survey on Deep Graph Representation Learning
• Adaptive Propagation Graph Convolutional Network
• Reconstruction Enhanced Multi-View Contrastive Learning for Anomaly Detection on Attributed Networks
• MGDCF: Distance Learning via Markov Graph Diffusion for Neural Collaborative Filtering
• SE-GSL: A General and Effective Graph Structure Learning Framework through Structural Entropy Optimization
• Learning Strong Graph Neural Networks with Weak Information
• X-GOAL: Multiplex Heterogeneous Graph Prototypical Contrastive Learning
• ROD: Reception-aware Online Distillation for Sparse Graphs
• Structure-Aware DropEdge Towards Deep Graph Convolutional Networks
• CLDG: Contrastive Learning on Dynamic Graphs
• Relational Self-Supervised Learning on Graphs
• Towards Complex Dynamic Physics System Simulation with Graph Neural ODEs
• DGDNN: Decoupled Graph Diffusion Neural Network for Stock Movement Prediction
• Learning Robust Representation through Graph Adversarial Contrastive Learning
• The Impact of Global Structural Information in Graph Neural Networks Applications
• Clarify Confused Nodes via Separated Learning
• Beyond Homophily with Graph Echo State Networks
• Scalable Deep Graph Clustering with Random-walk based Self-supervised Learning
• Contributions to Representation Learning with Graph Autoencoders and Applications to Music Recommendation
• Data-Driven Self-Supervised Graph Representation Learning
• On Inductive Biases for Machine Learning in Data Constrained Settings