Learning Dynamic Graph Representations

Rakshit Trivedi, Mehrdad Farajtabar, Prasenjeet Biswal, Hongyuan Zha

Abstract: We address two fundamental questions that arise in learning over dynamic graphs: (i) How to elegantly model dynamical processes over graphs? (ii) How to leverage such a model to effectively encode evolving graph information into low-dimensional representations? We present DyRep - a novel modeling framework for dynamic graphs that posits representation learning as a latent mediation process bridging two observed processes -- dynamic of the network (topological evolution) and dynamic on the network (activities of the nodes). To this end, we propose an inductive framework comprising of two-time scale deep temporal point process model parameterized by a temporal-attentive representation network and trained end-to-end using an efficient unsupervised procedure. We demonstrate that DyRep significantly outperforms state-of-art baselines for dynamic link prediction and event time prediction and provide extensive qualitative analysis of our framework.

Keywords: Dynamic Graphs, Representation Learning, Dynamic Processes, Temporal Point Process, Attention, Latent Representation

TL;DR: Models Representation Learning over dynamic graphs as latent mediation process bridging two observed processes of Topological Evolution of and Interactions on dynamic graphs.