A Creativity Survey of Unsupervised Graph Neural Network: Interactive Clustering and Embedding

摘要: The rise and application of neural network has successfully promoted the research of pattern recognition and data mining. In recent years, graph neural network has attracted more and more attention. It has some applications in text classification, sequence annotation, neural machine translation, relation extraction, image classification and other fields. This review mainly integrates the existing research on semi-supervised or unsupervised graph neural network. The research work of this paper is mainly classified in three aspects, one is based on the classification of research questions, the other is based on the classification of research methods, and the third is based on the classification of measures .The main research problems are the low-dimensional representation of nodes in graphs and the over-smooth problem in the process of message transfer. The research methods mainly focus on the graph embedding algorithm, such as the graph embedding algorithm based on probability graph and the method based on deep learning. The measurement methods mainly focus on the accuracy and efficiency of the algorithm and model .Finally, this paper also puts forward the feasible future research direction, which provides reference for readers.