Deep Learning Based Cryptographic Primitive Classification

Automated cryptographic classification framework using Intel's Pin platform for dynamic binary instrumentation and PyTorch for deep learning.

• Clone Repository
• Required Python libraries: sudo apt-get install python-pip python-tk
• Install requirements: pip install -r requirements.txt
• Install toolkit: python knight.py --setup
• Binary compilation requires OpenSSL: sudo apt install libssl-dev

Automatically draw distribution:

python crypto.py -d scale

Evaluatation:

python knight.py --predict <executable>
python knight.py --evaluate <dataset>

To add custom cryptographic samples to the generation pool, please follow the Format Specification.

We also published "CryptoKnight: Generating and Modelling Compiled Cryptographic Primitives " that can be found here in Open Access.

If you want to cite the paper please use the following format;

@Article{info9090231,
AUTHOR = {Hill, Gregory and Bellekens, Xavier},
TITLE = {CryptoKnight: Generating and Modelling Compiled Cryptographic Primitives},
JOURNAL = {Information},
VOLUME = {9},
YEAR = {2018},
NUMBER = {9},
ARTICLE NUMBER = {231},
URL = {http://www.mdpi.com/2078-2489/9/9/231},
ISSN = {2078-2489},
DOI = {10.3390/info9090231}
}