Mastering Machine Learning for Penetration Testing

EvadeML

EvadeML (https://evademl.org ) is an evolutionary framework based on genetic programming, for automatically finding variants that evade detection by machine learning based malware classifiers. It was developed by the Machine Learning Group and the Security Research Group at the University of Virginia.

To download EvadeML, clone it from https://github.com/uvasrg/EvadeML.

To install EvadeML, you need to install these required tools:

• A modified version of pdfrw for parsing PDFs: https://github.com/mzweilin/pdfrw
• Cuckoo Sandbox v1.2, as the oracle: https://github.com/cuckoosandbox/cuckoo/releases/tag/1.2
• The target classifier PDFrate-Mimicus: https://github.com/srndic/mimicus
• The target classifier Hidost: https://github.com/srndic/hidost ...