Autowaves in neuron-like medium
V. Yakhno
IAP RAS, 46 Uljanov str., 603600, Nizhny Novgorod, Russia
This paper deals with the investigation of possible dynamic structures in hierarchic distributed systems of coupled active elements.
Basic models of homogeneous neuron-like media and the corresponding sets of spatio-temporal solutions are considered. The self-similarity of architecture and functional regimes are shown at different levels of the hierarchic system. The sets of characteristic solutions for the collective activity structures in the models of neuron-like nonequilibrium media are considered. These solutions were used to demonstrate the procedures of extracting simplified images, followed by computing a necessary set of features.
Version of the model in which the neuron activity synchronization regimes in the cortex and in the thalamus occur because of the disturbance of the normal transformation of visual information when it is coded in the brain were considered.
The typical schemes for architecture of multilevel decision making systems are considered on the basis of the results obtained by the group working with IAP RAS. We believe that the algorithms fragment-by-fragment description is one of the universal approaches to encoding a complex image. There is an analogy with the mechanism used in the vision system of animals. In the same way the attention is focused on isolated fragments of a complex image, and there is similarity in the formation of the encoding trajectory between points of attention. Versions of a person access system using a complex images of biological objects are developed to verify our designing method.
This work was supported in part by the Russian Foundation of Basic Research under Grant No 95-02-04936- .