Impact of Solution Representation in Nature-Inspired Algorithms for Feature Selection

Abstract

This paper investigates how does the solution representation in nature-inspired algorithms impact the performance of feature selection in classification problems. Four most suitable nature-inspired algorithms for feature selection were considered in the analysis, namely the Differential Evolution, Artificial Bee Colony, Particle Swarm Optimization, and Genetic Algorithm. The binary-coded and real-coded variants of the mentioned algorithms were compared for filter-based and wrapper-based feature selection methodologies on datasets commonly used by the research community. Additionally, the algorithms' performance on reducing the feature subset size regarding different solution representations was compared. Statistical tests were performed for discovering any significant differences in the algorithms' performances.

Toward text psychology analysis using social spider optimization algorithm

“…These methods act independently of the learning algorithm therefor; they have a very high speed. Since they are not interacting with a learning algorithm, they are not very accurate 37 . (b) Wrapper‐based methods: In these methods, a learning algorithm is used to evaluate the set of features, and a feature set will be selected as the best feature set with the best accuracy and the least error.…”

“…(b) Wrapper‐based methods: In these methods, a learning algorithm is used to evaluate the set of features, and a feature set will be selected as the best feature set with the best accuracy and the least error. These methods have a relatively lower speed than those of the first group because the learning algorithm must be trained at each stage of the assessment, which is a very time‐consuming process, but they have a high accuracy 37 . So far, many methods have been proposed to solve the feature selection problem.…”