Date of Award

7-29-2020

Publication Type

Master Thesis

Degree Name

M.Sc.

Department

Computer Science

First Advisor

Sherif Saad Ahmed

Keywords

adaptive algorithms, anomaly detection, biometrics, keystroke biometrics, machine learning

Rights

info:eu-repo/semantics/embargoedAccess

Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Abstract

In this thesis, we argue that most of the work in the literature on behavioural-based biometric systems using AI and machine learning is immature and unreliable. Our analysis and experimental results show that designing reliable behavioural-based biometric systems requires a systematic and complicated process. We first discuss the limitation in existing work and the use of conventional machine learning methods. We use the biometric zoos theory to demonstrate the challenge of designing reliable behavioural-based biometric systems. Then, we outline the common problems in engineering reliable biometric systems. In particular, we focus on the need for novelty detection machine learning models and adaptive machine learning algorithms. We provide a systematic approach to design and build reliable behavioural-based biometric systems. In our study, we apply the proposed approach to keystroke dynamics. Keystroke dynamics is behavioural-based biometric that identify individuals by measuring their unique typing behaviours on physical or soft keyboards. Our study shows that it is possible to design reliable behavioral-based biometrics and address the gaps in the literature.

Available for download on Thursday, July 29, 2021

Share

COinS