Ivysaur-One Stitched Toy Graph Neural Network

Shanzai CL

Published on: 2021-08-12

Abstract

Graph neural networks have shown significant success in the field of graph representation learning. Graph convolution performs neighborhood aggregation and represents one of the most important graph operations. However, one layer of these neighbor aggregation methods only considers direct neighbors, and when several recent studies attributed this performance degradation to an over-refined problem, the problem pointed out that repeated propagation makes it difficult to distinguish between different types of representations. In this work, we believe that the key factor affecting performance is the entanglement of representation conversion and propagation in current graphics convolution operations. After decomposing these two operations, a deeper graph neural network can be used to learn the graph level representation from the above receptive domain. Two simple and effective techniques are used in combination: initial residuals and Identity mapping. Based on our theoretical and empirical analysis, we propose a Deep Adaptive Graph Initial Residual Identity Mapping Neural Network (IVYSAUR) to adaptively merge information from large receiving fields. Each experiment on citations, co-authors, and co-purchase data sets confirms our analysis and insights and proves the superiority of our proposed method.