Registration Number
0213U002153,
0110U000114
, R & D reports
Title
Research self-organizing treelike neural frames.
popup.stage_title
Head
Kovalevskii Sergey V.,
Registration Date
16-01-2013
Organization
Donbass State Engineering Academy
popup.description2
Research object: treelike neural networks. Purpose: further development of theory of neural networks technologies in area of design of the difficult systems and objects. Theoretical and practical results: the features of tree cascade of neural networks; the method of transformation of co-ordinates is investigational in part of his application at tuning of cascade treelike neural networks; going is developed near the topology placing of neural elements in the output cards of treelike cascade neural networks for their practical application; the ways of the practical application of neural networks, cascading tree. Scientific novelty: developed an algorithm for the formation of tree-like cascade of neural networks; the methods of synthesis and teaching of treelike cascade neural networks are developed; the method of normalization of basic data got further development for neural networks; a new method of formation of cellular automata. Efficiency: the offered developments allow improving quality of design of objects and prognostication of their indexes. Application domain: objects, described as difficult systems. The features of tree-like neural networks are considered a pre-treatment data for tree neural networks developed principles of synthesis tree cascade of neural networks, the coordinate transformation method is analyzed in terms of its application to set the cascade of neural networks are considered practical application of cascade neural networks for modeling complex systems and objects of different nature, as well as for the analysis of multivariate data.
Product Description
popup.authors
Гітіс Веніамін Борисович
Ковалевський Сергій Вадимович
Стародубцев Іван Миколайович
popup.nrat_date
2020-04-02
