Convolutional Neural Networks (CNNs) are artificial deep learning networks widely used in computer vision and image recognition for their highly efficient capability of extracting input image features. In the literature, such a successful tool has been leveraged for detection/classification purposes in several application domains where input data are converted into images. In this work, we consider the application of CNN models, developed by employing standard Python libraries, to detect and then classify Android-based malware applications. Different models are tested, even in combination with machine learning-based classifiers, with respect to two datasets of 5000 applications each. To emphasize the adequacy of the various CNN implementations, several performance metrics are considered, as also stressed by a comprehensive comparison with related work.
Image-based detection and classification of Android malware through CNN models
Aldini, Alessandro
;
2024
Abstract
Convolutional Neural Networks (CNNs) are artificial deep learning networks widely used in computer vision and image recognition for their highly efficient capability of extracting input image features. In the literature, such a successful tool has been leveraged for detection/classification purposes in several application domains where input data are converted into images. In this work, we consider the application of CNN models, developed by employing standard Python libraries, to detect and then classify Android-based malware applications. Different models are tested, even in combination with machine learning-based classifiers, with respect to two datasets of 5000 applications each. To emphasize the adequacy of the various CNN implementations, several performance metrics are considered, as also stressed by a comprehensive comparison with related work.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
