A security on neural cryptography with Multiple transfer functions using feedback And multi feedback tree parity machine

Abstract

A common secret key is generated using neural networks and newlineCryptography This can be achieved by two Tree Parity Machines TPMs newlinewhich are trained on their mutual output bits The random numbers are newlinegenerated by Tree Parity Machine TPM using Pseudo Random Number newlineGenerators PRNGs Then Tree Parity Machine s hidden layer output vectors newlineare compared That is output vectors of hidden unit using Hebbian learning newlinerule left dynamic hidden unit using Random walk learning rule and rightdynamic newlinehidden unit using Anti Hebbian learning rule are compared Among newlinethe compared values one of the values is received by the output layer In this newlinethesis the most successful attack on neural cryptography is the Majority newlineFlipping Attack which is presented here The probability of a successful newlineattack is calculated for different model parameters using numerical simulations newlineAlso the effective numbers of keys are increased using the entropy of the newlineweight distribution against brute force attack The genetic attack geometric newlineattack and majority attack are also explained in this thesis newline