SHA-AMD: sample-efficient hyper-tuned approach for detection and identification of Android malware family and category. (19th November 2021)

• Record Type: Journal Article
• Authors: Rasool, Aamir; Javed, Abdul Rehman; Jalil, Zunera
• Is Part Of: International journal of ad hoc and ubiquitous computing. Volume 38:Number 1/3(2021)
• Page Start: 172
• Page End: 183
• Publication Date: 2021-11-19
• Subjects: smart cities -- smart security -- Android malware -- family -- category -- deep learning -- machine learning -- cyber-attack; Ubiquitous computing -- Periodicals; Embedded computer systems -- Periodicals; Electronic data processing -- Distributed processing -- Periodicals; Wireless communication systems -- Periodicals; Computer architecture -- Periodicals; 004.2
• Languages: English
• ISSNs: 1743-8225

Abstract

Smart cities offer smart security solutions against cyber-attacks to the communities. Android-based smart devices have emerged as the best-selling artefact in the market. Due to this popularity and all-embracing usage, the Android operating system (OS) has become a lucrative target for attackers. In this paper, we propose an approach for the detection and classification of Android malware. We utilise ensemble machine learning (EML) comprising of support vector machines (SVM), decision tree (DT), K-nearest neighbours (KNN), and long short-term memory (LSTM) deep learning model for malware detection and classification in combination with feature selection to augment model performance. We present a comparison of EML, LSTM, and an ensemble of deep learning models: LSTM, gated recurrent unit (GRU), and recurrent neural network (RNN). Experimental results demonstrate better performance than state-of-the-art techniques using LSTM for binary classification at static layers and EML for category and family classification at dynamic layers.