Identification and Detection of Behavior Based Malware using Machine Learning

Malware that exploits the Web on a regular basis becomes a real menace. The transmission of malwareis very rapid during the last two decades which needs to bedetected. One of the efficient approaches for the detection of malware is manual heuristics analysis. To recognize and identification of behavior -based malware detection, machine learning techniques are in the consideration which may provide the optimum solution. Every malware's activity in a mimicked scenario would be continuously assessed, as well as behavior evaluations would be generated. Therefore, a major challenge is to identify the real-time identification of malware using signature matching algorithms. The classification methods used for this article such as KNN, J48, Decision Tree, SVM, Naïve Bayes, Neural Network, and Multilayer perceptron. J48 decision trees, which focused on the results of the testing that consisted of all 5 classifications, had a recall of 95.7%, a specificity of 97.3 percent, a positive predictive rate 2.4%,and accuracy is 96.8%. This was their collective best performance. In conclusion, malware could be identified inan efficient and effective way using a proof-of-concept that focuses on autonomous behavior-based malware classification.