Enhanced Android Malware Detection through Artificial Neural Networks Technique

Android devices are rapidly being used, which makes it easy for the malware threat to rise to higher levels. This ever-growing problem has prompted the need to enhance detection systems as far as these devices are concerned. Standard techniques of machine learning (ML) are sufficient from the point of view of their speed for searching patterns and behaviors of contemporary malware; however, it is more important to have effectively enhanced methods. The purpose of this paper is to expand the utilization of Android malware identification via artificial neural networks (ANNs) and compare its efficiency with that of other ML methods. An ANN is used in this study, and the results are compared against those of several other types of ML, including logistic regression (LR), k-nearest neighbors (KNN), extremely randomized trees (extra trees), gradient boosting (GBM), adaptive boosting (AdaBoost), and categorical boosting (CatBoost). The six evaluation values include training accuracy, testing accuracy, average accuracy, precision, recall and the F1 score. The ANN model performed well, with training and testing accuracies of 0. 99 and an average accuracy of 0. 99, precision of 0. 99, recall of 0. 98, whereas the F1 score, which is an average of both precision and recall, was 0. 99. Related studies based on conventional ML methods are also highly efficient, with some accuracy and an F1 score of 0. 95 and 0. 96. On the other hand, the ANN model ranked the best in the assessed measures. Thus, this study focuses on the reliability of ANNs for improving mobile security systems against next-generation malware and their applicability to secured Android smartphones. The reason behind the high accuracy of the ANN model is the enhanced learning ability of the ANN, which helps it learn the characteristics and dynamics of malware better than traditional ML models do.