Publicado 12 números por año
ISSN Imprimir: 1064-2315
ISSN En Línea: 2163-9337
Indexed in
Asymptotics of Linear Recurrent Regression under Diffuse Initialization
SINOPSIS
The behavior of recurrent least squares method in the absence of a priori information with respect to estimated unknown parameters is studied. The developed approach allows one to present its characteristics in analytical form, to explain the divergence phenomenon and to suggest a limiting recurrent estimation algorithm independent of a large parameter characterizing initial uncertainty and leading to divergence.
-
Skorohod B. A., Learning Algorithms for Neural Networks and Neuro-Fuzzy Systems with Separable Structures, Cybernetics and Systems Analysis, 51, 2, 2015. Crossref
-
References, in Diffuse Algorithms for Neural and Neuro-Fuzzy Networks, 2017. Crossref
-
Skorohod B., RLS Algorithms with Nonlinear Constraints and their Application to Reducing Estimates Fluctuations, 2019 International Russian Automation Conference (RusAutoCon), 2019. Crossref