Graph U-Nets

TL;DR

This paper introduces graph U-Nets with novel pooling and unpooling operations, enabling effective encoder-decoder architectures for graph data, and demonstrates improved performance on classification tasks.

Contribution

The paper proposes the first graph U-Net architecture with trainable pooling and unpooling layers tailored for graph data, addressing the challenge of graph pooling.

Findings

Achieves better accuracy than previous models on node classification.

Demonstrates effective graph segmentation with the proposed architecture.

Provides a novel approach to graph pooling and unpooling operations.

Abstract

We consider the problem of representation learning for graph data. Convolutional neural networks can naturally operate on images, but have significant challenges in dealing with graph data. Given images are special cases of graphs with nodes lie on 2D lattices, graph embedding tasks have a natural correspondence with image pixel-wise prediction tasks such as segmentation. While encoder-decoder architectures like U-Nets have been successfully applied on many image pixel-wise prediction tasks, similar methods are lacking for graph data. This is due to the fact that pooling and up-sampling operations are not natural on graph data. To address these challenges, we propose novel graph pooling (gPool) and unpooling (gUnpool) operations in this work. The gPool layer adaptively selects some nodes to form a smaller graph based on their scalar projection values on a trainable projection vector. We further propose the gUnpool layer as the inverse operation of the gPool layer. The gUnpool layer restores the graph into its original structure using the position information of nodes selected in the corresponding gPool layer. Based on our proposed gPool and gUnpool layers, we develop an encoder-decoder model on graph, known as the graph U-Nets. Our experimental results on node classification and graph classification tasks demonstrate that our methods achieve consistently better performance than previous models.