Enhancing Cryptanalysis of DES Encryption using Neural Networks and Firefly Algorithms
Varshini Balaji, Vallidevi Krishnamurthy · 2023
In today’s cryptographic landscape, evaluating encryption algorithms’ integrity is crucial. Neural networks offer promise for extracting information from Data Encryption Standard (DES) encrypted data. This report explores diverse methods of DES attacks using deep learning techniques. The primary aim is to decode DES-encrypted data with neural networks, extending this to contemporary encryption algorithms. The proposed system uses Long Short-Term Memory (LSTM) networks to improve plaintext reconstruction from DES-encrypted data. It evaluates LSTM network performance metrics and seeks to predict critical patterns. Findings reveal the neural network’s impressive ability to predict plaintext with 98% accuracy after 250k training iterations. However, performance varies with the encryption key, and complete plaintext recreation remains challenging. This method also experiments using the Firefly algorithm to enhance key prediction, improving overall efficiency. Artificial Neural Networks (ANNs) for classification tasks on encrypted data show promise for practical applications. Future work can focus on enhancing classification accuracy, exploring diverse data types, and studying the critical impact on performance metrics.