Comparative Study of Model Selection Methods for Networks

• Grant number: 25/04228-3
• Support Opportunities: Scholarships in Brazil - Master
• Start date: August 01, 2025
• End date: July 31, 2027
• Field of knowledge: Physical Sciences and Mathematics - Probability and Statistics - Statistics
• Principal Investigator: Andressa Cerqueira
• Grantee: Felipe Baptistão Durante Molina
• Host Institution: Centro de Ciências Exatas e de Tecnologia (CCET). Universidade Federal de São Carlos (UFSCAR). São Carlos , SP, Brazil
• Associated research grant: 23/13453-5 - Stochastic systems modeling, AP.TEM

Abstract

In recent years, complex networks have been widely used to model systems with objects of interest, called vertices, which interact with each other by a characteristic or relationship, represented by an edge. Through the various uses of complex networks, community detection has a great importance in the field, aiming to group vertices into the same cluster based on a given similarity between them. Thus, several models have been proposed to perform the clustering, assuming the number of communities is known previously, but this assumption is not always acceptable in real-world data. Therefore, in this work, we address the case of an unknown number of communities, that is a model selection problem, where a study will be conducted to compare selection methods for the Stochastic Block Model, including the proposed method, which is computationally approximated based on the normalized conditional maximum likelihood.