An approach to analysis of malware using supervised learning classification

In this, we propose an approach for improved malware class recognition with the reduced false positive rate. The technique developed here uses all frequency resolutions of Discrete Wavelet Transform this gives the information of texture features present in malware image with different resolutions based on the grayscale image. The efficiency of malware class identification or recognition is based on the wavelet transform. In this, paper we are using db4 wavelet family for decompose of malware into three level decomposing. The most effective statistical features like energy and standard deviations are calculated to build feature vector. The combined feature vector is used to identify the given malware samples which family it belongs. We are classifying the malware with more accuracy compared to existing work. The accuracy of malware class recognition is improved by 98.88% accuracy and precision is 99% respectively.