Self-Attention Graph Pooling

17 Apr 2019  ·  Junhyun Lee, Inyeop Lee, Jaewoo Kang ·

Advanced methods of applying deep learning to structured data such as graphs have been proposed in recent years. In particular, studies have focused on generalizing convolutional neural networks to graph data, which includes redefining the convolution and the downsampling (pooling) operations for graphs. The method of generalizing the convolution operation to graphs has been proven to improve performance and is widely used. However, the method of applying downsampling to graphs is still difficult to perform and has room for improvement. In this paper, we propose a graph pooling method based on self-attention. Self-attention using graph convolution allows our pooling method to consider both node features and graph topology. To ensure a fair comparison, the same training procedures and model architectures were used for the existing pooling methods and our method. The experimental results demonstrate that our method achieves superior graph classification performance on the benchmark datasets using a reasonable number of parameters.

PDF Abstract

Datasets


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Graph Classification D&D SAGPool_h Accuracy 76.45% # 30
Graph Classification D&D SAGPool_g Accuracy 76.19% # 32
Graph Classification FRANKENSTEIN SAGPool_g Accuracy 62.57 # 4
Graph Classification FRANKENSTEIN SAGPool_h Accuracy 61.73 # 5
Graph Classification NCI1 SAGPool_g Accuracy 74.06% # 35
Graph Classification NCI1 SAGPool_h Accuracy 67.45% # 46
Graph Classification NCI109 SAGPool_h Accuracy 67.86 # 20
Graph Classification NCI109 SAGPool_g Accuracy 74.06 # 16
Graph Classification PROTEINS SAGPool_h Accuracy 71.86% # 73
Graph Classification PROTEINS SAGPool_g Accuracy 70.04% # 77

Methods