Comparative Study: Mouth Brooding Fish (MBF) as a Novel Approach for Android Malware Detection

Abstract

Android Malware Detection has become increasingly prevalent, with the highest market share among all other mobile operating systems due to its open-source nature and user-friendliness. This has resulted in an uncontrolled proliferation of malicious applications targeting the Android platform. Emerging trends of Android malware are employing highly sophisticated detection and analysis evasion techniques, rendering traditional signature-based detection methods less effective in identifying modern and unknown malware.Alternative approaches, such as Machine Learning methods, have emerged as leading solutions for timely zero-day anomaly detection. Ensemble learning, a common meta-approach in machine learning, seeks to improve predictive performance by amalgamating predictions from multiple models. This paper introduces an enhanced strategy, Mouth Brooding Fish (MBF), based on ensemble learning for Android Malware Detection (AMD). The findings are further compared with the outputs of various algorithms including Support Vector Machine (SVM), AdaBoost, Multilayer Perceptron (MLP), Gaussian Kernel (GK), and Random Forest (RF). Compared to the other selected models, MBF exhibits remarkable performance with an F-score of 98.57%, precision of 99.65%, sensitivity of 97.51%, and specificity of 97.51%. Thus, the significant novelty of this work lies in the accuracy and authenticity of the selected algorithms, demonstrating their superior performance overall.