Hands-On Artificial Intelligence for Cybersecurity

Detecting metamorphic malware with HMMs

The examples of algorithms applied to malware detection that have been shown so far were intended to automate some of the routine activities performed by malware analysts.

However, the analysis methodology on which they are based is essentially static malware analysis.

Many of the concrete cases of malware threats, however, are not easily identifiable with this method of analysis, as the malware developers have learned how to work around the detection techniques based on signatures.

It will therefore be necessary to adopt a different methodology to identify the malicious behavior of more advanced malware, and to this end, we will have to move to an approach based on dynamic malware analysis, combining ...