Autowave processes

Spatio-temporal phenomena in distributed neuron-like media have been studied in IAP RAS for a number of years. Neuron-like media are distributed systems or neural network architectures consisting of active elements with several stationary (or quasi-stationary) states with nonlocal spatial coupling between the nonequilibrium elements. Basic models of neuron-like systems describing dynamics of homogeneous systems as well as hierarchical neuron-like architectures used for development of complex image recognition systems are studied. Dynamics of spatio-temporal (autowave) structures is investigated. The obtained solutions are used for interpreting dynamic modes of normal and pathologic transformation of sensor signals in physiological experiments (fig. 1). Algorithms of signal processing belong to the class of parallel algorithms of information processing (fig. 2). Variants of decision-making systems with adaptive algorithms (fig. 3) have been elaborated.

Fig. 1. Example of spatial self-excited structures in distributed systems and formation of burst mode of epileptiform activity caused by evolution of various autowave processes in a model of a three-component neuron-like medium. The plot shows variation of neuronal activity in time at the central point of the observation region.

Fig. 2. Example of track diagram of transformation of input image into “code” description of image fragment for further analysis.

Fig. 3. System for automatic human identification by hand shape.
    Scope:
  • access to protected data base in a computer
  • identification of bank client when working with bank accounts
  • access to military or nuclear objects