Event Details

Presenter: Faith Okpongete
Supervisor:

Date: Mon, July 18, 2022
Time: 09:00:00 - 10:00:00
Place: via Zoom - please see link below

ABSTRACT

Join Zoom Meeting

Meeting ID: 898 3738 8478

Password: 922431

Abstract: Over the past couple of decades, ransomware attacks have increased significantly and that

calls for more aggressive efforts in building robust detection models to detect and reduce the

impact of the attacks. Once attacked, the malware takes over the victims' machines and files

by locking or encrypting them. These attacks have also led to huge global financial loss for

people, businesses, and governments of nations. The cybercriminals who perpetrate these

attacks always demand payment of some ransom in cryptocurrency. Presently, there are

three common methods for detecting these ransomware attacks viz static, dynamic, and hybrid

detections. Static detection is known to evade detection easily by cryptographic techniques

and that is why dynamic detection was adopted for this project. We trained and tested

offline a detection model using the ISOT Ransomware dataset and implemented the proposed

model as a standalone endpoint detector. The detector was deployed and evaluated online

using new samples from the wild, whereby Cuckoo Sandbox was used to execute and extract

the malware features during the experiment. The online evaluation confirmed the offline

performance results, which were very encouraging.