Search academic texts

Information × Registration Number 0304U001900, 0101U002358 , R & D reports Title Neural-network based methods for accuracy impro vement of distributed popup.stage_title Розробка нейромережевих методів підвищення точності прогнозу дрейфу сенсорів, архітектури прецизійної ієрархічної дистрибутивної мережі збору і обробки сенсорних даних.Розробка спеціалізованого програмного забезпечення верхнього рівня мережі Head Sachenko Anatolyi Alekseevich, Registration Date 06-05-2004 Organization Ternopil Academy of National Economy popup.description2 Object of research - Нейросетевые methods of increase of accuracy of distributive networks of gathering and processing of the touch data. The purpose of work is improvement of operational characteristics of distributive networks for the account нейросетевой corrections of drift from the beginning of operation.Theoretical results of the given research - it is offered two new нейросетевых a method of increase of reliability of gathering and processing of the touch data, effective at the limited educational samples. Practical results - it is proved three-level structure of a highly effective distributive network which expands two-level structures of known distributive networks with use for the current processing the touch data of additional universal unit of an average level with remote reprogramming that has allowed специализировать processing of the touch data on levels of a distributive network it agrees real time of calculation drift and replacements of mathematical model of drift and by that to increase speed of data processing less, than twice; it is developed structure, algorithms and program modules of a computer of the top level of a distributive network on technology a client-server, that will allow to take into account specificity of hardware of a distributive network by division of processes of study and use of a neural network after levels with respective computing capacity, to simplify controllers of processing of the touch data and to provide replacement of mathematical model of drift in real time5635 Product Description popup.authors popup.nrat_date 2020-04-02 Close