Graph UNets
[edit]
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:20832092, 2019.
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 pixelwise prediction tasks such as segmentation. While encoderdecoder architectures like UNets have been successfully applied on many image pixelwise prediction tasks, similar methods are lacking for graph data. This is due to the fact that pooling and upsampling 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 encoderdecoder model on graph, known as the graph UNets. Our experimental results on node classification and graph classification tasks demonstrate that our methods achieve consistently better performance than previous models.
Related Material


