A Machine Learning Approach For Android Malware Recognition In View Of The Conjuction Of Elements By Utilizing Stacking Classifier

Abstract

This research examines Android malware detection utilizing diverse datasets, including "CIC_MALDROID2020," "Drebin," and "Malgenome." The extensive API and Permission data available through these databases enables thorough study to be undertaken. Various machine learning models utilized for classification include Logistic Regression, support Vector machine, ok-Nearest neighbors (KNN), decision Tree, Random woodland, and a Stacking Classifier that incorporates Random forest, Multi-Layer Perceptron (MLP), and LightGBM. The project involves meticulous dataset instruction, model training, and performance assessment. The primary objective of this method is to develop highly powerful fashions for Android virus detection. This program enhances mobile security with the aid of refining our model outcomes, hence augmenting our ability to detect and mitigate Android malware threats. Researchers and specialists in cellular security will locate the project's outcomes to be exceedingly fine. This paper also provides a Stacking Classifier that enhances characteristic extraction by means of integrating the top-rated features from Random forest, Multi-Layer Perceptron (MLP), and LightGBM. This ensemble learning strategy significantly improves prediction accuracy. Moreover, we advanced a stylish Flask framework integrated with SQLite to provide secure registration, authentication, and user trying out. This simplified system enhances the task’s overall efficacy via facilitating user enter and enabling prediction retrieval.