COUNTERING MARSHALKO'S ATTACKS ON ARTIFICIAL NEURON NETWORKS BY INTRODUCING FALSE CONNECTIONS

Abstract

The aim of the work is to add the neural network emulator software and the masking part of the neuron to the neural network training process according to GOST R52633 standard for user authentication systems, for the possibility of increasing the protection of the user's personal cryptographic keys in the event of a Marshalko attack.

Keywords

• neural network emulator
• cryptographic key
• marshalko attack
• neural network training
• user authentication