Optimization of the quantile criterion for the convex loss function by a stochastic quasigradient algorithm

A stochastic quasigradient algorithm is suggested for solving the quantile optimization problem with a convex loss function. The algorithm is based on stochastic finite-difference approximations of gradients of the quantile function by using the order statistics. The algorithm convergence almost sur...

Full description

Saved in:
Bibliographic Details
Published inAnnals of operations research Vol. 200; no. 1; pp. 183 - 198
Main Authors Kibzun, Andrey, Matveev, Evgeniy
Format Journal Article
LanguageEnglish
Published Boston Springer US 01.11.2012
Springer Science + Business Media
Springer Nature B.V
Subjects
Algorithms
Approximation
Business and Management
Combinatorics
Convergence
Convex analysis
Finite difference method
Operations research
Operations Research/Decision Theory
Optimization
Quantiles
Statistics
Stochasticity
Theory of Computation
Stochastic quasigradient algorithm
Quasiconcavity of the probability measure
The Minkowski inequality
Quantile function
Order statistics
Convergence almost surely
Online AccessGet full text
ISSN0254-5330
1572-9338
DOI10.1007/s10479-011-0987-z

Cover

More Information
Summary:A stochastic quasigradient algorithm is suggested for solving the quantile optimization problem with a convex loss function. The algorithm is based on stochastic finite-difference approximations of gradients of the quantile function by using the order statistics. The algorithm convergence almost surely is proved.
Bibliography:SourceType-Scholarly Journals-1
ObjectType-Feature-1
content type line 14
ObjectType-Article-2
content type line 23
ISSN:0254-5330
1572-9338
DOI:10.1007/s10479-011-0987-z