Dimension Reduction For Classification Using Principal Component Analysis (PCA) To Detect Malicious Executables

Jamal, Zalifh (2017) Dimension Reduction For Classification Using Principal Component Analysis (PCA) To Detect Malicious Executables. Masters thesis, Universiti Teknikal Malaysia Melaka.

	Text (24 Pages) Dimension Reduction For Classification Using Principal Component Analysis (PCA) To Detect Malicious Executables - Zalifh Jamal - 24 Pages.pdf - Submitted Version Download (93kB)
	Text (Full Text) Dimension Reduction For Classification Using Principal Component Analysis (PCA) To Detect Malicious Executables.pdf - Submitted Version Restricted to Registered users only Download (919kB)

Abstract

The development of cyberspace is not only facilitate people's lives. It should also be in line with security awareness related to personal and enterprise systems. Estimates of the number of new malware in 2013 reached 600 million, and has grown rapidly in recent years. Malware can attack a wide variety of computing devices and mobile devices are no exception. The number of malware attacks this execution on a large scale. This is a big challenge for malware detector. There are several ways of classification that are used to verify the accuracy of the research. Most classifiers have too many combinations that are difficult to assess, change often (optimal) and should get a brief training period. This study is aimed at reducing high-dimensional vector space to a lower dimension, thus reducing the problem of lack of accuracy of results. This study used a new approach, namely the Principal Component Analysis (PCA). PCA will make a classification so that the process can be done automatically and efficiently. PCA can reduce the number of dimensions of space by extracting features that describe the data set so that data sets can be confirmed precisely as if the entire data set together. The purpose of this study to investigate the malware will be selected, reducing the dimensions of the model that will be used to detect malware and to validate the models to find a minimum set of data to detect malware data. In order to find the right combination of features and options classification, two different sets of selection criteria used by two machine learning classifier. Result classification was assessed using the True Positive Rate (TPR), the false negative rate (FNR) and the accuracy of the feature selection approaching or exceeding 95% accuracy.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Malware (Computer software), Principal component analysis program, Computer security
Subjects:	Q Science > Q Science (General) Q Science > QA Mathematics
Divisions:	Library > Tesis > FTMK
Depositing User:	Nor Aini Md. Jali
Date Deposited:	25 Apr 2018 09:16
Last Modified:	16 Feb 2022 12:33
URI:	http://eprints.utem.edu.my/id/eprint/20734
Statistic Details:	View Download Statistic

Actions (login required)

View Item