DESIGN AND IMPLEMENTATION OF A MACHINE LEARNING-BASED MALWARE CLASSIFICATION SYSTEM WITH AN AUDIO SIGNAL FEATURE ANALYSIS APPROACH
Malware attacks carried out by cybercriminals are becoming increasingly well-organized because of the huge financial rewards that are to be accrued. However this causes se- vere losses to individuals and companies both directly and indirectly. Static analysis prevention can detect known malwa...
Saved in:
| Main Author: | |
|---|---|
| Format: | Theses |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/71776 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Malware attacks carried out by cybercriminals are becoming increasingly well-organized
because of the huge financial rewards that are to be accrued. However this causes se-
vere losses to individuals and companies both directly and indirectly. Static analysis
prevention can detect known malware with high accuracy but cannot overcome po-
lymorphism and packaging techniques. Meanwhile, dynamic analysis shows a high
frequency of false positive results. It is for this reason that a new approach to detect
malware uses the feature representation of the audio signal by converting the PE binary
file to audio file format and extracting into the time domain and frequency domain. The
results of this study are able to classify malicious files and distinguish them from be-
nign files, as well as classify these files into the appropriate families. The test results
show the best accuracy of 98.80% in distinguishing malware classes or benign, and
92.53% accuracy in identifying malware family classes using the XGBoost algorithm. |
|---|