Application of Multiobjective Genetic Algorithms in the Feature Selection Problem

• Grant number: 09/12963-2
• Support Opportunities: Scholarships in Brazil - Master
• Start date: March 01, 2010
• End date: December 31, 2010
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Ana Carolina Lorena
• Grantee: Newton Spolaôr
• Host Institution: Centro de Matemática, Computação e Cognição (CMCC). Universidade Federal do ABC (UFABC). Ministério da Educação (Brasil). Santo André , SP, Brazil

Abstract

The occurrence of irrelevant and/or redundant attributes in databases may impair the performance of computational processes for knowledge discovery, which motivates the application of feature selection techniques. The combinatorial nature of this problem makes the use of heuristics methods such as genetic algorithms appropriate, in order to obtain or approximate optimal subset of attributes.In many applications of feature selection one wants to optimize conflicting goals, such as the predictive performance of a subset of attributes and the cardinality of that subset. These characteristics enable the formulation of the feature selection task as a multiobjective optimization problem.This project aims to study and propose a method involving the use of multiobjective genetic algorithms in the feature selection problem, given the recent progresses that have been achieved in correlate state-of-art work.

Scientific publications

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

SPOLAOR, NEWTON; LORENA, ANA CAROLINA; LEE, HUEI DIANA. Feature Selection via Pareto Multi-objective Genetic Algorithms. APPLIED ARTIFICIAL INTELLIGENCE, v. 31, n. 9-10, p. 764-791, 2017. (09/12963-2, 12/22608-8)