Open AccessTRAINING METHODS FOR NEURAL NETWORKS USED IN THE DECISIONMAKING BLOCK OF SIGNALING DETECTION TOOLS
Author(s) -
E. I. Dukhan,
G. F. Zakharkin,
A. E. Dukhan
Publication year - 2020
Publication title -
vestnik urfo. bezopasnostʹ v informacionnoj sfere
Language(s) - English
Resource type - Journals
eISSN - 2225-5443
pISSN - 2225-5435
DOI - 10.14529/secur200305
Subject(s) - artificial neural network , false alarm , computer science , block (permutation group theory) , alarm , function (biology) , artificial intelligence , machine learning , signal (programming language) , feature (linguistics) , process (computing) , value (mathematics) , constant false alarm rate , pattern recognition (psychology) , algorithm , data mining , mathematics , engineering , linguistics , philosophy , geometry , evolutionary biology , biology , programming language , aerospace engineering , operating system
The article deals with the issue of training a neural network when building algorithms for detecting violators in the decision-making block of modern detection tools. The feature of train-ing neural networks in modern detection tools is to change the loss function under consider-ation, which takes into account the possible damage from the implementation of errors of the first and second kind in alarm systems. Based on the criterion of minimum average risk, it is advisable to minimize the probability of a false alarm (error of the first kind) with a fixed value of the probability of missing the target (error of the second kind). A new expression is obtained for updating the weights of the neural network during training, based on minimizing the new loss function. The process of training a neural network on a representative dataset of calculated information signal realizations and interference modeling is shown on the example of distrib-uted magnetometric systems. It is proved that the recurrent neural network has high character-istics of detecting violators: for a given value of correct detection of 0,95, the probability of a false alarm was 5,9∙10–4.