10 October، 2022
Discussion of a master’s thesis in the College of Computer Science and Mathematics – Department of Mathematics entitled: A proposed feature selection method based on the filter-envelope technology and the support vector machine for high-dimensional data sets
Discussion of a master’s thesis in the College of Computer Science and Mathematics – Department of Mathematics entitled:
A proposed feature selection method based on the filter-envelope technology and the support vector machine for high-dimensional data setsIn continuation of the scientific research movement, and with the follow-up and presence of the Dean of the College of Computer Science and Mathematics, Prof. Dr. Duha Bashir Abdullah, the respected
It was discussed in the discussion room of the Faculty of Computer Science and Mathematics at the University of Mosul on Monday 10/10/2022
Master’s thesis by student Wafaa Qassem Hammadi Ahmed and under the supervision of Prof. Dr. Omar Saber Qassem.The study focused on the Statistical Dependence (SD) technique as one of the filtering methods in order to arrange the data according to its importance in influencing the accuracy of classification, and the Atom Search Optimization (BASO) Binary algorithm was used in order to obtain the important features and neglect the features. After converting the basic algorithm for atomic search from Continuous Space to Discrete Space, the SD_BASO model has been proposed, which combines statistical dependence technology and binary atomic search improvement algorithm in order to obtain advanced data processing represented in reducing the dimensions of this data and obtaining High classification accuracy.The study focused on the Support Vector Machine (SVM) technology as a basic classifier for the proposed algorithm or model SD_BASO after making several comparisons between it and the Perceptron Network (PNN) and the Back-Propagation Artificial Neural Network (BPNN), and three were used. Kinds of kernel functions for SVM technology (Linear, polynomial, RBF) and building a special model for selecting its parameters through the ASO atomic search optimization algorithm and based on the concept of parameter tuning parameter in order to get the best classification accuracy.The study aims to apply and test the proposed SD_BASO algorithm on three types of datasets (Leukemia, Prostate, Lung) and make a set of comparisons with the usual algorithms and methods. Classification results for the data.The discussion committee consists of:Assistant Professor Dr. Safwan Omar Hassoun – Chairman
Assistant Professor Dr. Horaz Nazim Jabbar/Kirkuk University/College of Science
Dr. Talal Fadel Hussein – member
Prof. Dr. Omar Saber Qassem – Member and Supervisor
