IMPROVED ARTIFICIAL NEURAL NETWORK THROUGH METAHEURISTIC METHODS AND ROUGH SET THEORY FOR MODERN MEDICAL DIAGNOSIS

Ambily Merlin Kuruvilla, Dr. Balaji N.V.
2021 Indian Journal of Computer Science and Engineering  
A novel meta-heuristic soft computing model with feature selection implemented using Rough set (RS) theory for the diagnosis of coronary artery disease in diabetes patients is proposed in this study. The binary classification method in multiclass classification problems is applied by the One Versus Rest approach (OVR) is incorporated. To avoid the redundancy problem, a mathematical approach known as rough-set theory (RS) is applied to identify the most significant features from the dataset. The
more » ... Artificial Neural Network with improved hidden layers is used as the classifier which is optimized through a metaheuristic population-based method, known as the Grasshopper Optimization Algorithm (GOA) with a single objective optimization approach integrated for improving the accuracy of the model. Mean Square Error (MSE) is taken as the objective function and the result shows that the accuracy of the model has been improved significantly from 89.1% to 95.25% after optimization.
doi:10.21817/indjcse/2021/v12i4/211204161 fatcat:jslugzvbm5f4xmixchohazgtre