An enhanced classifier for authentication in keystroke dynamics using experimental data
Other Title(s)
مصنف مطور للتحقق من الهوية في ديناميكية الكتابة على لوحة المفاتيح باستخدام البيانات التجريبية
Dissertant
Thesis advisor
Comitee Members
Viktorov, Oleg
al-Dawud, Ali Asad Ahmad
University
Middle East University
Faculty
Faculty of Information Technology
Department
Computer Science Department
University Country
Jordan
Degree
Master
Degree Date
2014
English Abstract
The problem of cyber-attacks on information systems and networks, for various illegal purposes, has become a major threat to society and individuals.
Computer hackers are using all possible means to get access to private data, or to destroy such data.
It has become necessary to improve computer security through more advanced access control mechanisms.
Recently the use of biometrics has been employed to strengthen access control through user authentication that is based on users‟ measureable features.
The work in this thesis examines the keystroke dynamics approach, as a biometric authentication scheme that does not require extra hardware.
The study is focused on enhancing an anomaly detector that is based on a statistical model of classifying the typing rhythm of a person who is trying to access a computer system, whether it is a genuine user or an imposter.
An anomaly detector model is proposed, which uses the median for each typing feature element of as the point of center to measure acceptance against, and a Distance-to-Median threshold value which gives the upper and lower limits for an acceptable feature element.
The proposed model is evaluated using a public benchmark dataset of 20,400 records of password typing time measurement, collected by the Biometrics Lab of Carnagei Melon University.
The proposed model achieves lowest error rates of False Acceptance and False Rejection, compared to previous results of using other models on the same dataset.
A prototype keystroke dynamics authentication tool is developed, based on the proposed model.
The tool has two parts: training, to collect typing data of a user and from that data generates a user identity template, and testing module, to be used as an authentication tool that uses the data collected during training.
A discussion of the results of analyzing the benchmark data, and the data collected using the proposed models are discussed, and conclusions and suggestion for future work are presented.
Main Subjects
Information Technology and Computer Science
No. of Pages
80
Table of Contents
Table of contents.
Abstract.
Abstract in Arabic.
Chapter One : Introduction.
Chapter Two : Biometrics and authentication with literature study.
Chapter Three : The proposed keystroke dynamics model.
Chapter Four : The discussion of results.
Chapter Five : Conclusions and future work.
References.
American Psychological Association (APA)
al-Rahmani, Abd Allah Usamah. (2014). An enhanced classifier for authentication in keystroke dynamics using experimental data. (Master's theses Theses and Dissertations Master). Middle East University, Jordan
https://search.emarefa.net/detail/BIM-694056
Modern Language Association (MLA)
al-Rahmani, Abd Allah Usamah. An enhanced classifier for authentication in keystroke dynamics using experimental data. (Master's theses Theses and Dissertations Master). Middle East University. (2014).
https://search.emarefa.net/detail/BIM-694056
American Medical Association (AMA)
al-Rahmani, Abd Allah Usamah. (2014). An enhanced classifier for authentication in keystroke dynamics using experimental data. (Master's theses Theses and Dissertations Master). Middle East University, Jordan
https://search.emarefa.net/detail/BIM-694056
Language
English
Data Type
Arab Theses
Record ID
BIM-694056
