Approach for Detection of PE Malwares using Ensemble Learning and Deep Learning

Abstract: Security breaches are very common where the safety of the users is put to threat. Hence it is necessary that a threat to the system is identified which can be done with the help of malware detection. In order to explore, infect, steal data or virtually behave as the attacker wants with the help of a file or code delivered through a network is known as a malware. A PE malware typically is a malware code which is propagated through a PE file downloaded on the device which may result in loss of information and replacement of such malicious codes. Such malware creators get away with it easily due to traditional methods of testing which are unreliable and time consuming. The current thesis aims to deploy a prototype that uses the concepts of feature extraction and use the Portable Executable file at a later stage. These features extracted are fed to algorithms based on ML (machine learning) and deep learning so that the overall system of the model is enhanced when the feature undergoes layers of neutral networks. The model undergoes pre-processing techniques which is then fed to algorithms for training.

Keywords: PE files, malware, machine learning, deep learning