Section: New Software and Platforms

Behavioral Malware Analysis

Keywords: Artificial intelligence - Malware - Automatic Learning - Concolic Execution

Functional Description: Our approach is based on artificial intelligence. We extract graphs from programs, that represent their behaviors. Such graphs are called system call dependency graphs (SCDGs). Our software learns to distinguish malware from cleanware on a large set of malwares and cleanwares. Whenever we want to analyze a new program, we extract its graphs and use the result of the training to decide whether the new program to analyze is a malware.

• Partner: Cisco

• Contact: Axel Legay

• URL: https://team.inria.fr/tamis/