A Proposed Architecture Based on CNN for Feature Selection and Classification of Android Malwares

摘要: Malware detection process is based principally on extracting data given to classifier model; those are information about application’s behavior during its execution, permissions required by it or activities made in response some commands. Which clearly make the features chosen and build as vector highly influence credibility of model classifying with high accuracy unknown applications. For this reason, research field gave a decent attention resolve problematic malware models improving quality used classification process, performing feature selection processes order reduce dimensionality vectors, selecting most relevant, correlated informative eliminate redundant information. Many solutions were invented for purpose using machine-learning algorithm evaluate performance specific set filter algorithms that give rank each depending occurrence frequency, weight correlation. In paper, we proposed an approach CNN deep learning detecting android malwares solution redundancy problematic.