A Scalable Multiclass Algorithm for Node Classification

TL;DR

This paper presents MUCCA, a scalable, linear-time multiclass node classification algorithm based on game theory, which efficiently finds Nash Equilibria on spanning trees, maintaining accuracy while significantly improving speed.

Contribution

Introduces MUCCA, a novel linear-time multiclass node classification algorithm using game theory and spanning trees for scalability.

Findings

MUCCA is significantly faster than existing methods.

MUCCA achieves comparable predictive accuracy.

Experiments on real data validate its efficiency and effectiveness.

Abstract

We introduce a scalable algorithm, MUCCA, for multiclass node classification in weighted graphs. Unlike previously proposed methods for the same task, MUCCA works in time linear in the number of nodes. Our approach is based on a game-theoretic formulation of the problem in which the test labels are expressed as a Nash Equilibrium of a certain game. However, in order to achieve scalability, we find the equilibrium on a spanning tree of the original graph. Experiments on real-world data reveal that MUCCA is much faster than its competitors while achieving a similar predictive performance.