DeepGCNs: Can GCNs Go as Deep as CNNs?

TL;DR

This paper demonstrates that by adapting CNN techniques like residual connections and dilated convolutions, very deep GCNs can be trained effectively, leading to significant performance improvements in point cloud segmentation.

Contribution

Introduces methods to train very deep GCNs using CNN-inspired techniques, enabling models up to 56 layers deep with improved accuracy.

Findings

Deep GCNs outperform shallow models in segmentation tasks.

Residual and dilated connections mitigate vanishing gradient issues.

A 56-layer GCN achieves +3.7% mIoU over state-of-the-art.

Abstract

Convolutional Neural Networks (CNNs) achieve impressive performance in a wide variety of fields. Their success benefited from a massive boost when very deep CNN models were able to be reliably trained. Despite their merits, CNNs fail to properly address problems with non-Euclidean data. To overcome this challenge, Graph Convolutional Networks (GCNs) build graphs to represent non-Euclidean data, borrow concepts from CNNs, and apply them in training. GCNs show promising results, but they are usually limited to very shallow models due to the vanishing gradient problem. As a result, most state-of-the-art GCN models are no deeper than 3 or 4 layers. In this work, we present new ways to successfully train very deep GCNs. We do this by borrowing concepts from CNNs, specifically residual/dense connections and dilated convolutions, and adapting them to GCN architectures. Extensive experiments show the positive effect of these deep GCN frameworks. Finally, we use these new concepts to build a very deep 56-layer GCN, and show how it significantly boosts performance (+3.7% mIoU over state-of-the-art) in the task of point cloud semantic segmentation. We believe that the community can greatly benefit from this work, as it opens up many opportunities for advancing GCN-based research.