engleski

Discrete time neural network syntesis using interaction activation functions

A new structure of descrete time neural networks is proposed. For this purpose the following concepts are used: (i) an introduction of an interaction activation functions, (ii) input time-varying signal distribution, (iii) time discrete domain synthesis, and (iiii) one step learning iteration procedure. Simulation results, based on the synthesis of a new structure of a recurrent neural network of an universal logical unit are presented. The proposal neural network synthesis precedure is useful for applications to identification and control of nonlinear dynamic systems, like industrial robots.

Neural network; discrete time synthesis; interaction activation functions

nije evidentirano