Hybrid Multimodal Machine Learning Driven Android Malware Recognition and Classification Model

Android mobile applications and devices can be extremely utilized and employed in industry and smart cities. Malware recognition is an efficient and powerful method for assurance security of Android schemes, mainly for industrial platforms and smart cities. In recent times, research utilizing machine learning (ML)-based approaches for Android malware recognition enhanced quickly. However, one of the performed methods to accomplish feature selection and analysis, so-called feature engineering is labour-intensive and depends on artificial knowledge. Therefore, this study presents a Hybrid Multimodal Machine Learning Driven Android Malware Recognition and Classification (HM3-AMRC) model. The proposed HM3-AMRC technique recognizes and categorizes the Android malware accurately and efficiently. In the presented HM3-AMRC technique, initial phase of data preprocessing is performed. Next, the classification process takes place using three ML techniques such as gradient boosting decision tree (GBDT), extreme learning machine (ELM), and multilayer perceptron (MLP). Moreover, Bayesian optimization (BO) technique is employed for optimal parameter selection of the ML approaches. The stimulation analysis of the HM3-AMRC method is tested on Android malware dataset. Extensive comparison analysis highlighted the betterment of the HM3-AMRC method over existing techniques.