Алгоритмы использования нейронных сетей для улучшения системы RSA

Abstract

This study explores the application of neural network algorithms to improve the security and efficiency of the RSA cryptosystem. With the advent of quantum computing and advanced cryptanalysis techniques, traditional RSA implementations face increasing vulnerabilities. We investigate the use of Generative Adversarial Networks (GANs) for key generation, Long Short-Term Memory (LSTM) networks for encryption optimization, and Convolutional Neural Networks (CNNs) for attack detection. Experimental results demonstrate that GANs enhance key randomness, LSTMs reduce encryption time by 15 %, and CNNs achieve 98 % accuracy in attack detection. The findings highlight the potential of neural networks to reinforce RSA against modern threats while maintaining computational efficiency