Parçacık Sürü Optimizasyonu Yoluyla Geliştirilen Doğrusal Bir Sınıflandırıcının Analizi

Meta-heuristics are high-level approaches developed to discover a heuristic that provides reasonable solution many varieties of optimization problems. The classification problems contain sort problem. Simply, the objective herein is reduce number misclassified instances. In this paper, question whether meta-heuristic methods can be used construct linear models or not answered. To end, Particle Swarm Optimization (PSO) has been engaged address Classifier (PSC) with certain function compared Support Vector Machine (SVM), Perceptron Learning Rule (PLR), and Logistic Regression (LR) applied fifteen data sets. experimental results point out PSC compete other classifiers, it turns superior classifiers for some binary Furthermore, average accuracies PSC, SVM, LR, PLR 80.8%, 80.6%, 80.9%, 57.7%, respectively. order enhance performance more advanced functions developed. Further, accuracy boosted by constructing tighter constraints via another meta-heuristic.