Simplifying Graph Convolutional Networks
arXiv:1902.07153
Abstract
Graph Convolutional Networks (GCNs) and their variants have experienced significant attention and have become the de facto methods for learning graph representations. GCNs derive inspiration primarily from recent deep learning approaches, and as a result, may inherit unnecessary complexity and redundant computation. In this paper, we reduce this excess complexity through successively removing nonlinearities and collapsing weight matrices between consecutive layers. We theoretically analyze the resulting linear model and show that it corresponds to a fixed low-pass filter followed by a linear classifier. Notably, our experimental evaluation demonstrates that these simplifications do not negatively impact accuracy in many downstream applications. Moreover, the resulting model scales to larger datasets, is naturally interpretable, and yields up to two orders of magnitude speedup over FastGCN.
In ICML 2019. Code available at https://github.com/Tiiiger/SGC
References in corpus (1)
Cited by in corpus (9)
- Fast Graph Representation Learning with PyTorch Geometric
- Revisiting Graph Neural Networks: All We Have is Low-Pass Filters
- Simple BERT Models for Relation Extraction and Semantic Role Labeling
- Adversarial Examples on Graph Data: Deep Insights into Attack and Defense
- Meta-GNN: On Few-shot Node Classification in Graph Meta-learning
- Graph Star Net for Generalized Multi-Task Learning
- PAN: Path Integral Based Convolution for Deep Graph Neural Networks
- Fisher-Bures Adversary Graph Convolutional Networks
- Redundancy-Free Computation Graphs for Graph Neural Networks