Multi-Objective Optimization through Objective Space Partitioning

• Grant number: 25/13944-4
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: December 01, 2025
• End date: November 30, 2026
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Alexandre Cláudio Botazzo Delbem
• Grantee: Luis Eduardo Aires Coimbra
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Associated research grant: 24/08485-8 - Center for artificial intelligence in health management, AP.PCPE

Abstract

This Scientific Initiation project proposes the investigation of the (Multi-Objective Evolutionary Algorithm with Many Tables - MEAMT) algorithm. This method has shown potential for handling many-objective problems, i.e., with three or more objectives. Its principle involves exploring subpopulations associated with each objective individually, as well as all combinations of objectives. In this project, a review of the main evolutionary algorithms will be carried out, with emphasis on those with potential to deal with many objectives. Furthermore, improvements to MEAMT will be investigated with the aim of increasing its capacity to address many-objective problems. The central hypothesis is that different forms of objective separation into subspaces affect the convergence and diversity of the generated solutions, especially in many-objective problems. Performance analyses will be conducted using standard benchmark functions from the literature. The expected advancements in this project could be applied in the future to solve complex health management problems, where multiple conflicting objectives need to be balanced, such as resource allocation, service planning, and performance indicator improvement. (AU)