Uma Versão Binária da Meta-heurística Water Flow Optimizer Aplicada à Seleção de Características

Abstract

In this work, a binary version of the Water Flow Optimizer (WFO) algorithm, called Binary Water Flow Optimizer (BWFO), is introduced addressing the feature selection problem. WFO is an evolutionary algorithm inspired by the way water flows in nature. In this new approach, the BWFO uses the laminar flow and turbulent flow operators in a binary version, using the accuracy of the Optimum-Path Forest (OPF) classifier as an objective function. The proposed approach is evaluated through a comparative analysis made with classical methods of dimensionality reduction, more specifically with the Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA), as well as the metaheuristics Binary Water Wave Optimization (BWWO) and Binary Bat Algorithm (BBA). The computational results demonstrate that the approach is a valid and effective alternative for feature selection problems.