Learning to Generate Networks

TL;DR

This paper explores the capability of various deep generative models to learn and replicate the complex structures of networks from data, highlighting limitations in modeling larger networks.

Contribution

It evaluates the effectiveness of deep belief networks, dependency networks, and exponential random graph models in generating networks that match input data, revealing size-related limitations.

Findings

Deep models can capture small network behaviors.

Models struggle with larger networks beyond a few nodes.

No current model fully captures complex behaviors in larger networks.

Abstract

We investigate the problem of learning to generate complex networks from data. Specifically, we consider whether deep belief networks, dependency networks, and members of the exponential random graph family can learn to generate networks whose complex behavior is consistent with a set of input examples. We find that the deep model is able to capture the complex behavior of small networks, but that no model is able capture this behavior for networks with more than a handful of nodes.