Malware detection based on string length histogram using machine learning

• Published in: 2018 3rd IEEE International Conference on Recent Trends in Electronics, Information & Communication Technology (RTEICT) pp. 1836 - 1841
• Main Authors: Sawaisarje, Snehalkumar K., Pachghare, Vinod K., Kshirsagar, Deepak D.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-05-2018
• Subjects: Feature extraction; Grippers; histogram; Histograms; Machine learning; machine learning classifiers; malware detection; printable string length; Support vector machines; Trojan horses
• DOI: 10.1109/RTEICT42901.2018.9012242

In the past decade, the creation of new malicious programs has been on the rapid rise [1], with thousands of malware released each day. The challenge is to analyze such a huge number in a timely efficient manner and detect malicious program accurately. As the nature of the malware keeps changing, any solution should be flexible enough to easily adapt to these changes. The paper proposes malware detection approach based on frequency distribution of length of the printable strings for efficient detection of malware such as worm and backdoor. This method is fast, simple and easily adaptable to new types of malware. The proposed malware detection approach consists of static binary file scanning, feature extraction, machine learning classifiers and malware detection. The proposed implemented system tested on VX heavens dataset which provides the highest accuracy of 89% for decision tree classifier as compared to 79.14% accuracy of KNN.