Machine learning for the prediction of voter model opinions through complex network structures

TL;DR

This paper demonstrates that machine learning can accurately classify opinions in the voter model using topological features from complex networks, extending to multiple opinion states and highlighting key network metrics.

Contribution

It introduces a generalized machine learning approach for predicting voter model opinions based on network topology, applicable to various dynamical processes on complex networks.

Findings

High accuracy in classifying opinions with topological features

Key network metrics identified and ranked by importance

Method applicable to multiple opinion states in voter models

Abstract

The inference of outcomes in dynamic processes from structural features of systems is a crucial endeavor in network science. Recent research has suggested a machine learning-based approach for the interpretation of dynamic patterns emerging in complex network systems. The hypothesis is applied in this study towards showing opinions can be classified in the voter model with the use of topological features from complex networks. An analysis was first performed with 2 (0 or 1) opinions from a voter model and extended for 3 and then 4 opinions from the same model. A breakdown of the key network topological features for the estimation is provided and network metrics are ranked in the order of importance with high accuracy. Our generalized approach is applicable to dynamical processes running on complex networks. This study is a step towards the application of machine learning methods to studies of dynamical patterns from complex network systems.