|
Analysis of the convergence of the algorithm for constructing a convex regression dependence
A. A. Gudkov, S. P. Sidorov, K. A. Spiridonov Saratov State University
Abstract:
In this paper, to solve the problem of constructing a convex approximation to noisy data, we propose an algorithm for constructing convex regression using the active set approach. It is shown that the algorithm converges to the optimal solution and it is found the estimate of its complexity.
Keywords:
nonlinear optimization, monotonic regression, convex regression, active set, segment regression.
Citation:
A. A. Gudkov, S. P. Sidorov, K. A. Spiridonov, “Analysis of the convergence of the algorithm for constructing a convex regression dependence”, Proceedings of the 20 International Saratov Winter School "Contemporary Problems of Function Theory and Their Applications", Saratov, January 28 — February 1, 2020. Part 2, Itogi Nauki i Tekhniki. Sovrem. Mat. Pril. Temat. Obz., 200, VINITI, Moscow, 2021, 115–125
Linking options:
https://www.mathnet.ru/eng/into907 https://www.mathnet.ru/eng/into/v200/p115
|
Statistics & downloads: |
Abstract page: | 117 | Full-text PDF : | 67 | References: | 18 |
|