|edoc-Server der Humboldt-Universität zu Berlin|
Warren Powell, Princeton University|
Andrzej Ruszczynski, Rutgers University
Huseyin Topaloglu, Cornell University
|Title:||Learning algorithms for separable approximations of stochastic optimization problems|
|Date of Acceptance:||05.11.2002|
Stochastic Programming E-Print Series |
|Editors:||Julie L. Higle; Werner Römisch; Surrajeet Sen|
|Metadata export: To export the complete metadata set as Endote or Bibtex format please click to the appropriate link.||Endnote Bibtex|
|print on demand: If you click on this icon you can order a print copy of this publication.|
|We propose the use of sequences of separable, piecewise linear approximations for solving classes of nondiffferential stochastic optimization problems. The approximations are estimated adaptively using a combination of stochastic gradient information and possibly sample information on the objective function itself. We prove the convergence of several versions of such methods when the objective function is separable and illustrate their behavior on numerical examples. We then demonstrate the preformance on nonseparable problems that arise in the context of two-stage stochastic programming problems, and demonstrate that these techniques provide near optimal solutions with a very fast rate of convergence compared to other solution techniques.|
These data concerning access statistics for individual documents
have been compiled using the webserver log files aggregated by AWSTATS.
They refer to a monthly access count to the full text documents as well as to the entry page.
As for format versions of a document which consist of multiple files (such as HTML) the highest monthly access number to one of the files (chapters) is shown respectivly.
To see the detailled access numbers please move the mouse pointer over the single bars of the digaram.
Gesamtzahl der Zugriffe seit Jul 2011: