Skip to content

Latest commit

 

History

History
7 lines (6 loc) · 576 Bytes

README.md

File metadata and controls

7 lines (6 loc) · 576 Bytes

MalConv-DOS_PERTURBATION

Python implementation of the misclassification attack on MalConv Neural Network, it is based on Explaining Vulnerabilities of Deep Learning to Adversarial Malware Binaries, proposed by Demetrio et al.

Usage

It simply requires a binary sample as argument and it will return the DOS header before and after the attack, so:
python attack.py <binary>
At the end of interations a graph of model accuracy during iterations will be provided.