HOME JOURNALS CONTACT

Journal of Engineering and Applied Sciences

Machine Learning Based Key Generating for Cryptography
Hayfaa A. Atee, Robiah Ahmad, Norliza Mohd Noor and Abidulkarim K. Ilijan

Abstract: An efficient key generation technique is demanding for a greatly secured cryptosystem. Traditional key generation techniques are very systematic which makes it subject to attack easily. The inclusion of Artificial Neural Network (ANN) algorithm in the cryptosystem is found to enhance the cryptographic performance in terms security and robustness to attack. Based on Extreme Learning Machine (ELM) for one hidden layer NN, we propose a sub-key generation approach for achieving a good cryptosystem. To initialize the input-hidden layer weights and data in each round, the initial key has been designed to include the ANN topology, activation function and seeds for Pseudo Random Number Generator (PRNG). The sub-key in each round is created using the output layer weights. Evaluation measures of the developed approach demonstrated complete sensitivity and predictability. Furthermore, the achieved remarkable reduction in the risks of breaking the symmetric key algorithm is attributed to the generation of independent sub-key in each round. Our secured sub-key generation approach may contribute towards the development of a secured cryptographic system.

How to cite this article
Hayfaa A. Atee, Robiah Ahmad, Norliza Mohd Noor and Abidulkarim K. Ilijan, 2016. Machine Learning Based Key Generating for Cryptography. Journal of Engineering and Applied Sciences, 11: 1829-1834.

© Medwell Journals. All Rights Reserved