Patterns of functional activity of neuron-like cognitive systems
Using up-to-date methods of nonlinear dynamics and neurophysiology data the IAP RAS resear-chers constructed basic models of neuron-like systems describing dynamic modes of functional activity that are adequate to available experimental data on the features of living system perception and response to various sensor signals. Models with biology-inspired architecture are used for creation of technical devices (simulators) capable of reproducing basic regularities of the behavior of living systems. The following results were obtained at the Laboratory of Autowave Processes.
Modeling processes of information signal transformation in neuron-like systems
Biometric systems for human recognition. A functional model of the processes of sensor signal processing in thalamocortical brain networks was developed. Such universal systems are capable of transforming sensor signals of different modalities. Modes of normal information processing and system self-excitation were demonstrated on an example of the interaction of four neuronal modules each of which simulates processes in neural ensembles of the cortex, cortical interneurons, neurons of thalamic sensor nuclei and thalamic reticular nuclei (I. V. Nuidel, M. E. Sokolov, V. G. Yakhno).
A basic modification of shock simulator stably reproducing a typical mode of interaction of three neuroendocrinal systems was developed based on data on neurochemical mechanisms of extreme states (three-component theory of stress and shock). The dynamics of physiological reactions recorded in laboratory experiments and in clinical studies under analogous actions was successfully reproduced by means of a mathematical model. Dynamic modes were classified and the principle settings of the basic simulator module were determined. Further research will allow developing the information technology that would optimize diagnostics of the course of extreme states and elaboration of antishock drugs (S. B. Parin, S. A. Polevaya, V. G. Yakhno).
An autonomous integral biometric recognition system (IBRS) was developed that demonstrates the operating capacity of the basic models of neuron-like systems for automatic verification and identification of people by their biometric parameters such as face, hand, fingerprint images, and voice sounds. The presented IBRS modality was tested on the basis of data for 150–200 persons. The recognition error was within 0.1–1% and slightly increased, depending on the conditions of teaching and identification, with increasing number of users (A. A. Telnykh, A. V. Kovalchuk, O. V. Shemagina, V. G. Yakhno).
Study of diagnostic features for assessing the functional state of living systems
Mathematical models of collective dynamics of neuronal populations of the sensor zones of cerebral cortex with parameters consistent with data of psychophysical experiment were developed. The neural network based on modified pulsed neurons describes impulse activity distribution in two symmetric (right- and left-hemispheric) neuronal populations containing 100 elements each under the action of dichotic stimulus.
A system for personified monitoring and remote diagnostics of human functional state under the conditions of free behavior was developed based on the integration of WEB technologies, wireless sensor networks and miniature sensor platforms. A unique opportunity was created for acquiring knowledge about natural dynamics of the interaction оf human physiological systems responsible for recognition, attention, memory, decision making, emotions, organization of behavior without distance and mobility limitations. Analysis of data monitoring by the methods of nonlinear dynamics permits personified mapping of stressor, continuous control of extreme and optimal functioning of the organism, and early diagnostics of dangerous functional states.
Study of the dynamics of molecular self-assembly processes in multicomponent liquids
A new methodo-logical approach to assessing the quality of multicomponent liquids without determining their composition was proposed (Т. А. Yakhno, А. G. Sanin, О. А. Sanina, V. G. Yakhno). The approach is based on the phenomenon of self-organization of drops drying on a solid wetted substrate. The physicochemical mechanisms of the self-organization processes were studied. A device for registering the dynamics of mechanical and acoustic properties of drying drops representing the dynamics of self-assembly of liquid components and their transition to solid state was developed. Complex physicochemical processes in a drying drop are integrally plotted as a curve of acoustomechanical impedance (AMI). It was shown experimentally that the shape of the AMI curve is a fingerprint of the liquid. The cognitive component of this approach is reduction of the dimension of measurement space by registering one integral parameter – AMI dynamics. Quantitative comparison of the shape of the curves of the studied liquids with the reference ones permits establishing authenticity of the liquids. This simple, fast and cheap method was successfully tested for medical and veterinary express-diagnostics for revealing fake beverage and drugs. It is quite possible that an analogous mechanism will work in the case of ordor perception in living organisms.