Network classification through random walks

doi:10.1016/j.chaos.2025.116817

Network classification through random walks

2025-07-24

0

PRE

AI

Odemir Martinez Bruno *

DOI：10.1016/j.chaos.2025.116817

Original

Original request for help

Save

Abstract

En 中文

• Proposes a novel set of network features based on random walk statistics. • Achieves top classification accuracy in 10 out of 12 datasets. • Demonstrates strong robustness to structural noise. • Highlights self-avoiding walk visit stats as a lightweight, efficient classifier. • Validated across diverse domains (synthetic, metabolic, bioinformatics, social).

Keywords:

random walk statistics

network features

classification accuracy

self-avoiding walk

robustness to noise

Network classification through random walks

Abstract

Journal

Organization