An Efficient Antimalware System Using Neural Networks with Support Vector Machine and Discriminative Common Vector

There is a plenty of malicious software available in internet. Internet users are in high risk in downloading such malicious software. The existing antivirus packages lacks in generalization ability. So an efficient antimalware system was proposed using Neural Networks. The antimalware system is treated as a classification problem. There are two approaches considered for the classification viz., classifying the malware using MLP with Support Vector Machine and by using Radial Basis Function neural network with Discriminative Common Vector (DCV). The malware binaries are first converted into grey scale images and then the images are classified using MLP-SVM and RBF-DCV networks. Based on the results it was found that RBF with DCV network results in better score for classifying malware images with 89.16%. The next stage towards building an anti-malware system shall be to explore other better neural network models for classification.