B1810
Title: Intensity profile projection: A framework for continuous-time representation learning for dynamic networks
Authors: Patrick Rubin-delanchy - University of Bristol (United Kingdom) [presenting]
Abstract: A new algorithmic framework, intensity profile projection, is presented for learning continuous-time representations of the nodes of a dynamic network, characterised by a node-set and a collection of instantaneous interaction events which occur in continuous time. The framework consists of three stages: estimating the intensity functions underlying the interactions between pairs of nodes, e.g. via kernel smoothing; learning a projection which minimises a notion of intensity reconstruction error; and inductively constructing evolving node representations via the learned projection. It is shown that the representations preserve the underlying structure of the network, and are temporally coherent, meaning that node representations can be meaningfully compared at different points in time. Estimation theory is developed which elucidates the role of smoothing as a bias-variance trade-off, and shows how smoothing can be reduced as the signal-to-noise ratio increases on account of the algorithm `borrowing strength' across the network.
