This repository contains the final project for the Machine Learning for Cyber-Security course at New York University Tandon School of Engineering, in collaboration with Naman Patel.
Malicious software in the form of computer viruses, Trojan horses, bots, and Internet worms like adware, spyware, and ransomware poses a serious threat to computer security. The amount of different malwares and its possible variants are numerous which makes classical condition based or signature based approaches ineffective. Although these malicious software are plentiful, these variants of malware families share typical behavioral patterns reflecting its origin and purpose. In this paper we study the capability of various language modeling based approaches to extract these behavioral pattern for system call based malware detection. A detailed analysis of the effectiveness of various language modeling based features, namely, Bag of Words, Term Frequency - Inverse Document Frequency and word representation is presented along with their performance using classifiers based on Naïve Bayes, SVM and logistic regression on the MALREC dataset.