Progressive Graph Convolutional Networks for Semi-Supervised Node Classification

Graph convolutional networks have been successful in addressing graph-based\ntasks such as semi-supervised node classification. Existing methods use a\nnetwork structure defined by the user based on experimentation with fixed\nnumber of layers and neurons per layer and employ a layer-wise propagation rule\nto obtain the node embeddings. Designing an automatic process to define a\nproblem-dependant architecture for graph convolutional networks can greatly\nhelp to reduce the need for manual design of the structure of the model in the\ntraining process. In this paper, we propose a method to automatically build\ncompact and task-specific graph convolutional networks. Experimental results on\nwidely used publicly available datasets show that the proposed method\noutperforms related methods based on convolutional graph networks in terms of\nclassification performance and network compactness.\n