"Machine Learning for Cryptographic Algorithm Identification”

Abstract

This project explores the application of machine learning techniques for the identification of cryptographic algorithms based on various features extracted from ciphertext. The dataset consists of synthetic data with features such as byte frequency, Shannon entropy, block size, key length, ciphertext length, n-gram frequency, and statistical moments (mean, variance, skewness, kurtosis). These features are crucial for distinguishing between different cryptographic algorithms. The objective is to classify a set of cryptographic algorithms, including AES, DES, RSA, Blowfish, RC4, Twofish, and Triple DES (3DES), based on the given input features. Two powerful machine learning algorithms, Random Forest and XGBoost, are employed to model the classification process. The machine learning models are trained and tested using the synthetic data to predict the type of cryptographic algorithm used in encryption. The primary focus is on the accuracy and robustness of the models in identifying the correct algorithm based on statistical analysis of the ciphertext. This project provides an efficient way of classifying cryptographic algorithms without requiring access to the encryption keys, which can be valuable in situations involving ciphertext analysis.

Keywords: Machine Learning, Cryptographic Algorithm, Classification, Random Forest, XGBoost, AES, DES, RSA, Blowfish, Ciphertext.