年間 12 号発行
ISSN 印刷: 1064-2315
ISSN オンライン: 2163-9337
Indexed in
Study of Convergence of One-Step Adaptive Identification Algorithms
要約
The question of convergence of the regularized one-step adaptive algorithms of Kaczmarz and Nagumo–Noda, used for solving the identification problem, is studied. Estimates of the rate of algorithms convergence are obtained, and it is shown that the introduction of the regularization parameter together with improving the computational stability of the algorithms slightly retards the process of identification. The presence of information about statistical properties of useful signals and interferences enables the selection of the algorithm parameters providing their maximum convergence rate.
-
Rudenko Oleg G., Bezsonov Oleksandr O., ADALINE Robust Multistep Training Algorithm, Control Systems and Computers, 3 (287), 2020. Crossref