S.S. Antsyferov - Dr. Sc. (Eng.), Professor, Moscow Technological University (MIREA) E-mail: c_standard@fel.mirea.ru K.N. Fazilova - Post-graduate Student, Moscow Technological University (MIREA) E-mail: fazilova@mirea.ru K.E. Rusanov - Ph. D. (Eng.), Associate Professor, Moscow Technological University (MIREA) E-mail: c_standard@fel.mirea.ru

The development of artificial intelligence led to formation a new approach, which involves the creation of intelligent systems based on the neurophysiological principles of construction of the human nervous system, i.e., cognitive neural systems. The theoretical basis of this direction serve the biological model of the functioning of the nervous system, the theory of formal neurons, the dynamic model of neural networks, learning methods of associative networks, etc. In recent years, work is underway both on the properties and standardization of indicators of cognitive systems properties. One of the defining property is stability. It should be noted that the concept of stability of the cognitive system acquires a meaning different from conventional technical systems: they must function in the regime of non-equilibrium stability that enables rapid transformation of their structural construction. Resulting from the task to establish the nomenclature and normalized values of indicators of sustainability appear to be currently relevant. The aim of this work is to identify indicators of non-equilibrium stability of cognitive systems. These indicators are determined by the accepted model of cognitive systems. The most common model is the Hopfield model. According to this model, the description of the functioning of the neural network are variables that describe the state of neurons S, and variables describing the connections between neurons J.The stability condition is determined by the ratio of the number of key images М that can be saved among neurons in the network N. However, the Hopfield model does not reflect the purposeful process of transformation of the structure of systems when changing the content and the intensity of the input information flows. We assume that one of the conditions of non-equilibrium of cognitive systems is the active nature of the functioning of neural elements. An adequate model in this case can serve as a probabilistic Bayesian approach to the construction of algorithms of decision-making systems under conditions of variability and uncertainty. The proposed model of the functioning of cognitive systems with active elements allows to determine the main indicators of non-equilibrium stability: parameter λ characterizing the strength of interaction between neurons; system entropy H; function h(a), f(a), g(a) and their parameters (k1, k2, x - the base of the logarithm).

 

1. Ancyferov S.S. Formirovanie spektra teplovykh izobrazhenijj obektov i raspoznavanie ikh obrazov // Opticheskijj zhurnal. 1999. T. 66. № 12. S. 44−48.
2. Ancyferov S.S. Formirovanie spektra teplovykh izobrazhenijj obektov i raspoznavanie ikh obrazov // J. Opt. Technol. 1999. V. 66. № 12. P. 1047−1049.
3. Ancyferov S.S., Evtikhiev N.N., Golub B.I. Adaptacija informacionno-raspoznajushhikh biomedicinskikh sistem // Biomedicinskaja radioehlektronika. 2001. № 1. S. 5−9.
4. Ancyferov S.S., Evtikhiev N.N., Golub B.I. Tekhnologija adaptivnojj obrabotki informacii teplovykh shirokospektralnykh polejj // Naukoemkie tekhnologii. 2002. T. 3. № 4. S. 45−50.
5. Ancyferov S.S. Metrologija virtualnykh sistem // Izmeritelnaja tekhnika. 2003. № 5. S. 17−21.
6. Ancyferov S.S., Evtikhiev N.N. Adaptivnye sistemy raspoznavanija obrazov prostranstvenno-vremennykh polejj // Iskusstvennyjj intellekt. 2004. № 3. S. 405−416.
7. Ancyferov S.S., Evtikhiev N.N. Adaptivnaja obrabotka informacii prostranstvenno-vremennykh izotropnykh polejj // Opticheskijj zhurnal. 2006. T. 3. № 10. S. 52−57.
8. Ancyferov S.S. Obshhie principy postroenija i zakonomernosti funkcionirovanija intellektualnykh sistem // Iskusstvennyjj intellekt. 2011. № 3. S. 6−15.
9. Ancyferov S.S. Voprosy metrologicheskogo obespechenija intellektualnykh sistem // Mir izmerenijj. 2012. № 5. S. 46−51.
10. Ancyferov S.S. Metrologicheskie osnovy formirovanija informacionnogo prostranstva obrazov v intellektualnykh sistemakh obrabotki // Iskusstvennyjj intellekt. 2012. № 4(58). S. 283−291.
11. Ancyferov S.S., Rusanov K.E. Bystrodejjstvie intellektualnykh sistem // Iskusstvennyjj intellekt. 2013. № 4(62). S. 259−265.
12. Ancyferov S.S. Standartizacija pokazatelejj kachestva produkcii kognitivnykh tekhnologijj // Naukoemkie tekhnologii. 2014. T. 15. № 7. S. 7−13.
13. Ancyferov S.S. Problemy iskusstvennogo intellekta // Problemy iskusstvennogo intellekta. 2015. № 0(1). S. 5−12.