Classification of Encrypted Texts using Deep Learning

Abstract

The most widely used cryptographic systems can identify cryptographic algorithms and identify encryption keys. Statistical methods and learning a variety of machines have been used to identify cryptographic algorithms, each of which has its own advantages and disadvantages. This paper seeks to provide a method for identifying the algorithm used for encrypted texts in text files. Since the volume of this kind of data is very big and increases at any given moment, then the accuracy is calculated by voting of these classifiers. The process of identifying the encryption algorithm is also known from the encrypted texts as the classification of text. So, three methods of encryption AES, RC5, BLOWFISH have been used to evaluate system performance. A three class's classifier is needed, for this purpose, k-nearest neighbor's algorithm has been used. This article is based on a deep learning approach, provides a new method for identifying the pattern in cryptographic texts and learning them by methods of representing features. The proposed method, consists of four parts of the preprocessing, feature learning, data classification and voting. The proposed system's efficiency in algorithm classification is 99.1%.