A Neural Network is a machine that is designed to model the way in which the brain performs a task or function of interest. It has the ability to perform complex computations with ease. The objective of this project was to investigate the use of ANNs in various kinds of digital circuits as well as in the field of Cryptography.
During our project, we have studied different neural network architectures and training algorithms. A comparative study is done between different neural network architectures for an Adder and their merits/demerits are discussed. Using a Jordan (Recurrent network), trained by back-propagation algorithm, a finite state sequential machine was successfully implemented.
The sequential machine thus obtained was used for encryption with the starting key being the key for decryption process. Cryptography was also achieved by a chaotic neural network having its weights given by a chaotic sequence.