Search academic texts

Information × Registration Number 0204U005144, 0101U001762 , R & D reports Title Theoretical basis and mathematical support for neuro-fuzzy-systems of early diagnostics, forecasting and simulation in apriori and current uncertainty conditions elaborations popup.stage_title Head Bodyanskiy Ye.V., Registration Date 22-03-2004 Organization Kharkov national university of radioelectronics popup.description2 Intelligent methods of data processing are investigated. The goal is the development of architectures and learning algorithms for specialized neuro-fuzzy networks (both feedforward and recurrent) for the forecasting of non-stationary stochastic and chaotic sequences that can change their properties at a priori unknown moments, as well as methods of image processing in multi-dimensional feature spaces under prior and current uncertainty. Research methods are the methods of computational intelligence. New methods for the synthesis of artificial neural network architectures were developed. The networks comprise recurrent neurons and their learning algorithms with optimal performance. A neural network-based approach to the processing of stochastic and chaotic sequences was proposed. Recognition methods in multi-dimensional feature spaces under uncertainty were developed. There are no analogs for the results obtained. The results are fundamental and are a contribution to the general theory of computationalintelligence. Practical results lie in the possibility of application of the proposed systems to the real-time analysis of arbitrary time series, as well as to the valid decision making in mechatronics, process automation, processing of biological, economical, and financial indices, correlation-extremal navigation systems, graphical and graphological information processing systems, and medical diagnostic equipment. Product Description popup.authors popup.nrat_date 2020-04-02 Close