A Unified View on Graph Neural Networks as Graph Signal Denoising
Authors:
Open Access
- 26 October 2021
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 1202-1211
Abstract
Graph Neural Networks (GNNs) have risen to prominence in learning representations for graph structured data. A single GNN layer typically consists of a feature transformation and a feature aggregation operation. The former normally uses feed-forward networks to transform features, while the latter aggregates the transformed features over the graph. Numerous recent works have proposed GNN models with different designs in the aggregation operation. In this work, we establish mathematically that the aggregation processes in a group of representative GNN models including GCN, GAT, PPNP, and APPNP can be regarded as (approximately) solving a graph denoising problem with a smoothness assumption. Such a unified view across GNNs not only provides a new perspective to understand a variety of aggregation operations but also enables us to develop a unified graph neural network framework UGNN. To demonstrate its promising potential, we instantiate a novel GNN model, ADA-UGNN, derived from UGNN, to handle graphs with adaptive smoothness across nodes. Comprehensive experiments show the effectiveness of ADA-UGNN.Keywords
All Related Versions
Funding Information
- Army Research Office (W911NF-21-1-0198)
- NSF (National Science Foundation) (IIS1714741, CNS1815636, IIS1845081, IIS1907704, DRL2025244, IIS1928278, IIS1955285, IOS2107215, IOS2035472)
This publication has 7 references indexed in Scilit:
- DEMO-Net: Degree-specific Graph Neural Networks for Node and Graph Classification2019
- Large-Scale Learnable Graph Convolutional Networks2018
- Graph Convolutional Neural Networks for Web-Scale Recommender Systems2018
- Geometric Deep Learning on Graphs and Manifolds Using Mixture Model CNNs2017
- Label Informed Attributed Network Embedding2017
- Signal denoising on graphs via graph filtering2014
- Adaptive piecewise polynomial estimation via trend filteringThe Annals of Statistics, 2014