A Knowledge-Domain Analyser for Malware Classification
Om Prakash Samantray, Satya Narayan Tripathy · 2020 International Conference on Computer Science, Engineering and Applications (ICCSEA) · 2020
Malware or malicious software is a serious threat from decades. Different malware detection and analysis approaches have been developing, but new malware and evasion techniques are also generating with the same pace. In this paper we have proposed a framework called knowledge-domain mal-ware analyser which analyses and detects malware using machine learning classification techniques. To carry out our experiment, we have collected a data set from online sources which contains 100000 records and 35 attributes. Feature selection methods are performed before classification to include most relevant features for classification. We have used extra tree classifier and k-best feature selection method to select best and relevant features of the collected data set. Machine learning classification algorithms such as support vector machine, Logistic Regression and Naive Bayes are applied on the data set and a comparison of their detection accuracy is presented.