The Extension of Graph Convolutional Neural Network with Capsule Network for Graph Classification

Abstract

In this paper we extend a graph convolutional neural network (GCNNs) which is the one of the existing state-of-art deep learning methods using the notion of capsule networks for graph classification. Through experiments, we show that by extending GCNNs using capsule networks can significantly overcome the challenges of GCNNs for the task of graph classification.