End-to-End Learning from Complex Multigraphs with Latent-Graph Convolutional Networks

TL;DR

This paper introduces Latent-Graph Convolutional Networks (L-GCNs) for end-to-end learning on complex multigraphs with rich edge labels, demonstrating improved performance on synthetic and real-world datasets.

Contribution

The paper proposes L-GCNs, a novel model that propagates information through a latent adjacency tensor to handle complex multigraphs with rich edge information.

Findings

Nonlinear interactions per neighbor improve classification accuracy

Model performs well in inductive learning scenarios

Effective on both synthetic fraud detection and real urban transportation data

Abstract

We study the problem of end-to-end learning from complex multigraphs with potentially very large numbers of edges between two vertices, each edge labeled with rich information. Examples range from communication networks to flights between airports or financial transaction graphs. We propose Latent-Graph Convolutional Networks (L-GCNs), which propagate information from these complex edges to a latent adjacency tensor, after which further downstream tasks can be performed, such as node classification. We evaluate the performance of several variations of the model on two synthetic datasets simulating fraud in financial transaction networks, ensuring the model must make use of edge labels in order to achieve good classification performance. We find that allowing for nonlinear interactions on a per-neighbor basis boosts performance significantly, while showing promising results in an inductive setting. Finally, we demonstrate the use of L-GCNs on real-world data in the form of an urban transportation network.