A Tool For Malware And Malicious URL Classification (MAMUC)

Abstract

With the rise in use of internet ,also have grown the security concerns associated with it. The most common threats
that we encounter on the internet are Malicious URLs and Malware.Traditional solutions for combating these threats, are to
build a databases of known sources of trouble or/and using filters to restrict access to resources. This is however not dynamic
enough to detect attacks of smart Cyber-criminals. Techniques such as type-squatting, domain-squatting , code obfuscation,etc
are hard to identify using conventional methods. Thus to stay one step ahead in the war against Cyber-crime we need to use
Machine and Deep learning enabled methods in our defense equipment .
The tool Malware and Malicious URL Classifier (MAMUC) is a unique tool which has 3 features. The first feature helps us to
identify if a given URL is malicious or benign. This classifier runs on a novel Malben dateset. The second feature is used to
covert a malware sample from byte-code to its corresponding malware image. This called as malware visualization,which is a
strategy used in static malware inspection. The third feature of the tool is a malware classifier. Once malware is converted to
an image ,its analysis is simplified to a case of image classification. This part identifies to which family an input malware is
most closely associated with. Once the malware family is identified it is easier to deal with it. The novelty we have employed
in this classifier is the use of Dual channel CNN(DCCNN) algorithm for malware classification