Multi-layer network composition under a unified dynamical process

Abstract

In this paper, we take a step towards a principled method of network composition from multi-layer data. We argue that inter-layer dynamics is a essential component of understanding the structure as a whole. Mathematically, we consider the following abstract problem: given multiple layers of network data over a shared vertex set, and additional parameters for inter-layer transitions, construct a (single) weighted network that best integrates the multi-layer dynamics. In this context, we will also study an empirical use case of the composition framework.