1 AIT Asian Institute of Technology

GA and statistics-based feature selection algorithm

AuthorThanakorn Sakchaicharoenkul
Call NumberAIT Thesis no. CS-99-16
Subject(s)Genetic algorithms
NoteA thesis submitted in the partial fulfillment of the requirements for the degree of Master of Engineering, School of Engineering and Technology
PublisherAsian Institute of Technology
AbstractFeature selection is the method that used to reduce the redundant features. In this study, the methodology consists of several stages that are related to one another. Genetic Algorithm is applied as a feature selection procedure. Many of statistic knowledge are used to create new fitness functions and new classification procedures. The new fitness functions are based on coefficient of determination and partial correlation coefficient. The new classification procedures are based on Linear Regression Analysis and Linear Discriminant Analysis. The aim of this work is to make proposed feature selection procedure generate the better results, both remaining features and predicted accuracy, than the prior methods. And the learning period can be shortened. Finally, GA with Linear Discriminant and two proposed fitness functions can generate the better results.
Year1999
TypeThesis
SchoolSchool of Engineering and Technology (SET)
DepartmentDepartment of Information and Communications Technologies (DICT)
Academic Program/FoSComputer Science (CS)
Chairperson(s)Yulu, Qi
Examination Committee(s)Sadananda, Ramakoti;Phan Minh Dung
DegreeThesis (M.Eng.) - Asian Institute of Technology, 1999


Usage Metrics
View Detail0
Read PDF0
Download PDF0