An Efficient Anti-Malware System With Multi Layer Perceptron And Discriminative Common Vector

Abstract

In this current internet era security is one of the major concerns. There are lots of freely available software for download and many of them could be malicious to cause harm to the computer and network. There are new families of malwares developed and released on day-to-day basis. The existing anti-malware systems like anti-virus packages need to be updated frequently. There are chances that new malwares are not detected by the old packages. Hence there is a need for an efficient anti-malware system. This paper describes an alternate way to detect and classify malwares using Multi-Layer Perceptron (MLP). The malware are binary files. The binary files are converted into grey scale images. The converted images are classified using MLP-DCV. The result shows a classification accuracy of 93% when MLP is applied with DCV.