The threats impose by the cyber-attacks due to malicious software (malware) have been increasing drastically with the evolution of information technology. Since people use web applications on a daily basis these malware attacks have become challenging. There have been various attacks affecting confidentiality, integrity and availability of data which has become a major security concern. Though the manual inspection and classification methods seemed to bring up some light to this facet, these methods are no longer considered effective, since they are time consuming and inefficient. With the high-rate malware spreading, it is a necessity to come up with some novelty approach to classify them as malware or benign software. So, this is where machine learning comes up as a novelty approach in malware classification. In this paper, a malware dataset was used on several machine learning classifiers like Support Vector Machinery (SVM) and Gaussian Naive Bayes classifiers were used and Recurrent Neural Network (RNN) and Convolutional Neural Networks (CNN) were used as the deep learning classifiers. Although there are many other methods for malware classification, a machine learning approach could be efficient and effective in detecting malicious software. Thus, the primary objective of this paper is to provide an insight to the machine learning approach in malware classification by depicting, which is the best classifier of the listed, that can effectively classify malware based on their accuracy or precision. In conclusion, based on the results this recognizes Recurrent Neural networks as the best approach that recorded the highest accuracy.

Malware Classification Machine Learning Deep Learning Binary Classification