On building ensembles of diverse and competent classifiers

• Grant number: 22/10917-8
• Support Opportunities: Scholarships in Brazil - Doctorate (Direct)
• Start date: October 01, 2022
• End date: June 21, 2026
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Ana Carolina Lorena
• Grantee: Victor Castro Nacif de Faria
• Host Institution: Divisão de Ciência da Computação (IEC). Instituto Tecnológico de Aeronáutica (ITA). São José dos Campos , SP, Brazil
• Associated research grant: 21/06870-3 - Beyond algorithm selection: meta-learning for data and algorithm analysis and understanding, AP.JP2
• Associated scholarship(s): 23/11704-0 - Dynamic multiple predictive models for multi-view learning, BE.EP.DD

Abstract

Ensemble learning is a popular strategy to take advantage of classification techniques with distinct biases, by joining their competences and capabilities. This work will investigate strategies to generate ensembles of classifiers based on their competences as evaluated by a meta-learning strategy. Nonetheless, joining different classifiers is profitable only when their errors are uncorrelated, so that the errors of a set of classifiers can be compensated by the correct predictions of an- other set. Therefore, measures of diversity of the classifiers' predictions will also be considered. (AU)