Interacting neural networks and cryptography
Abstract
Description
Two neural networks which are trained on their mutual output bits are analysed using methods of statistical physics. The exact solution of the dynamics of the two weight vectors shows a novel phenomenon: The networks synchronize to a state with identical time dependent weights. Extending the models to multilayer networks with discrete weights, it is shown how synchronization by mutual learning can be applied to secret key exchange over a public channel.
Invited talk for the meeting of the German Physical Society
Invited talk for the meeting of the German Physical Society