Network representation learning: an overview

Abstract

Networks are important ways of representing objects and their relationships. A key problem in the study of networks is how to represent the network information properly. With the developments in machine learning, feature learning of network vertices has become an important area of study. Network representation learning algorithms turn network information into dense, low-dimensional real-valued vectors that can be used as inputs for existing machine learning algorithms. For example, the representation of vertices can be fed to a classifier such as a Support Vector Machine (SVM) for vertex classification. In addition, the representations can be used for visualization by taking the representations as points in a Euclidean space. The study of network representation learning has attracted the attention of many researchers. In this article, recent works on network representation learning are introduced and summarized.