Box-constrained optimization for minimax supervised learning***

Autor:	Lionel Fillatre, Susana Barbosa, Cyprien Gilet
Jazyk:	angličtina
Rok vydání:	2021
Předmět:	Mathematical optimization T57-57.97 Simplex Applied mathematics. Quantitative methods Discretization Computer science Supervised learning Probabilistic logic Constrained optimization MathematicsofComputing_NUMERICALANALYSIS Function (mathematics) Minimax QA1-939 Subgradient method Mathematics
Zdroj:	ESAIM: Proceedings and Surveys, Vol 71, Pp 101-113 (2021) HAL
ISSN:	2267-3059
Popis:	In this paper, we present the optimization procedure for computing the discrete boxconstrained minimax classifier introduced in [1, 2]. Our approach processes discrete or beforehand discretized features. A box-constrained region defines some bounds for each class proportion independently. The box-constrained minimax classifier is obtained from the computation of the least favorable prior which maximizes the minimum empirical risk of error over the box-constrained region. After studying the discrete empirical Bayes risk over the probabilistic simplex, we consider a projected subgradient algorithm which computes the prior maximizing this concave multivariate piecewise affine function over a polyhedral domain. The convergence of our algorithm is established.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_dedup___::cab2b48f04b8ec694a53394ec7ae038f https://www.esaim-proc.org/articles/proc/pdf/2021/02/proc2107109.pdf Zobrazit plný text záznamu Plný text ve formátu PDF