Representation Learning from Time Varying Networks and its Application to Temporal Link Prediction

Abstract

Machine learning with graph-structured data has gained broad research interest in newlinerecent years due to the increased importance of performing network mining tasks newlineon data from various domains. Generating efficient network representation is one newlineimportant challenge in applying machine learning over network data. Recently, newlinerepresentation learning methods are widely used in various domains to generate newlinelow dimensional latent features from complex high dimensional data. A significant amount of research effort is made in the past to generate node representations newlinefrom graph-structured data using representation learning methods. Most of these newlinemethods are only applicable to static networks and therefore cannot capture the newlineevolving nature and temporal dynamics of time-varying networks. This research newlineaims to develop representation learning methods for two different dimensions of newlinetime-varying networks, namely dynamic networks and temporal networks. newline