Link Prediction in Graphs with Autoregressive Features

TL;DR

This paper introduces a novel approach for link prediction in dynamic graphs by leveraging autoregressive features modeled via VAR, optimizing adjacency and VAR matrices jointly with efficient proximal algorithms.

Contribution

It proposes a joint optimization framework that incorporates sparsity and low-rank properties for improved link prediction in evolving graphs.

Findings

Derived oracle inequalities illustrating trade-offs in smoothing parameters.

Efficient computation using proximal methods and generalized forward-backward algorithm.

Enhanced prediction accuracy by modeling autoregressive features in dynamic graphs.

Abstract

In the paper, we consider the problem of link prediction in time-evolving graphs. We assume that certain graph features, such as the node degree, follow a vector autoregressive (VAR) model and we propose to use this information to improve the accuracy of prediction. Our strategy involves a joint optimization procedure over the space of adjacency matrices and VAR matrices which takes into account both sparsity and low rank properties of the matrices. Oracle inequalities are derived and illustrate the trade-offs in the choice of smoothing parameters when modeling the joint effect of sparsity and low rank property. The estimate is computed efficiently using proximal methods through a generalized forward-backward agorithm.