Date of Award

Fall 12-9-2021

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

School

Computing Sciences and Computer Engineering

Committee Chair

Andrew H. Sung

Committee Chair School

Computing Sciences and Computer Engineering

Committee Member 2

Beddhu Murali

Committee Member 2 School

Computing Sciences and Computer Engineering

Committee Member 3

Ahmed Sherif

Committee Member 3 School

Computing Sciences and Computer Engineering

Committee Member 4

Parthapratim Biswas

Committee Member 4 School

Mathematics and Natural Sciences

Committee Member 5

Sungwook Lee

Committee Member 5 School

Mathematics and Natural Sciences

Abstract

Rapid advances in artificial intelligence (AI), machine learning (ML), and deep learning (DL) over the past several decades have produced a variety of technologies and tools that, among numerous cybersecurity issues, have enticed cybercriminals and hackers to design malware for the Android operating systems and/or manipulate multimedia. For example, high-quality and realistic fake videos, images, or audios have been created to spread misinformation and propaganda, foment political discord and hate, or even harass and blackmail people; these manipulated, high-quality and realistic videos became known recently as Deepfake. There has been much work done in recent years on malware analysis and detection in Android applications, and many solutions have been suggested to cope with the issues highlighted by Deepfake. This dissertation addresses a couple of research topics: first, based on manifest analysis, it introduces a feature-based detection technique for detecting Android malware; second, it investigates DL and non-DL methods for detecting Deepfake videos. The first research shows that the obtained results outperform all published work in Android malware detection concerning the Drebin dataset. The outcomes of the second work demonstrate that the classical ML-based methods alone can obtain superior performance in the detection of Deepfake. In Android malware research, we propose a substring-based feature selection (SBFS) strategy and assess using various ML algorithms to identify Android malware. In addition, we apply ensemble-based learning techniques as well as advanced ensembled techniques. In Deepfake research, this study presents both DL and non-DL methods. For identifying Deepfakes, we propose a deep ensemble learning-based method called DeepfakeStack in a DL-based approach where an enhanced composite classifier is created by combining a set of current DL-based models. In the non-deep learning-based method, we use a traditional ML method based on conventional feature creation and feature selection approaches to train, tune and test ML classifiers. These two pieces of research can offer a promising basis for building effective systems for detecting the two cybersecurity threats: Android (Mobile) malware and Deepfake.

ORCID ID

https://orcid.org/0000-0001-6626-4189

Available for download on Friday, July 15, 2022

Share

COinS