Using the Capabilities of Artificial Neural Networks in the Cryptanalysis of Symmetric Lightweight Block Ciphers

Cryptanalysis is a scientific field that determines the cryptographic algorithm's secret key or stability level. Nowadays, there are various methods of cryptanalysis of modern cryptographic algorithms. However, deep learning-based cryptanalysis is being actively studied following the development of artificial intelligence technologies. Cryptanalysis techniques such as plaintext attacks on lightweight block cyphers are known today. In this paper, we consider a method of training an artificial neural network through a separate model for each key byte of deep learning. We use this method to predict the key of the lightweight block S-AES algorithm. And we propose a deep-learning cryptanalysis method based on each byte of the key. Using this method, the training accuracy of the neural network was 92.80%, and the validation accuracy was 53.30%.