The structure of the neurocomputer system classification of signals

Author(s):  N.I. Korsunov, Dr., Prof., Belgorod National Research University, Belgorod, Russia

S.N. Ushakova, Belgorod National Research University, Belgorod, Russia, ushakova_s@bsu.edu.ru

Issue:  Volume 46, № 3

Rubric:  Computer simulation history

Annotation:  The article proposes to use a neurocomputer system using two-dimensional binary maps of signal characteristics (the system is based on quantization of signals in amplitude and time) to improve the performance of signal recognition systems without the requirements of their reproduction after recognition. This occurs in connection with the use of neurocomputer systems implicit and explicit representation of time, which leads to a number of advantages and disadvantages. Neural networks with an implicit representation of time should have the property of dynamism, which is provided by the introduction of a network of direct propagation of time delays, and in the recognition of signals used neurocomputer systems with an explicit representation of time. Explains the construction of brain-computer system employing a two-dimensional binary map of characteristics of signals, and provides the framework that implements the method of classification of signals with explicit representation of time. This leads to an increase in the efficiency (speed) of the network.

Keywords:  neurocomputer system, neural network, signal recognition, signal classification.

Full text (PDF):  Download

Downloads count:  259