CALCULATION OF THE PROBABILITIES OF REPETITION OF COMMON CONNECTIONS FOR NEURONS WITH 4 AND 5 INPUTS AT RANDOM SELECTION OF 512 BIOMETRIC PARAMETERS FOR DATA OF COMPARABLE QUALITY

Abstract

This article discusses the attack of G.B. Marshalko aimed at simplifying the task of selecting a key for a neural network biometric converter into a code. Simulation modeling calculated the probabilities of occurrence of different starting conditions for the implementation of an attack on neural networks with 512 inputs. Networks with artificial neurons with 4 and 5 inputs have been studied.

Keywords

• neural networks
• biometrics
• simulation modeling
• information security