MEM pixel correlated solutions for generalized moment and interpolation problems

I. Csiszár, Fabrice Gamboa, Elisabeth Gassiat

Research output: Contribution to journalArticle

18 Citations (Scopus)

Abstract

In generalized moment problems (signed) measures are searched to fit given observations, or continuous functions are searched to fit given constraints. Known convex methods for solving such problems, and their stochastic interpretations via Maximum Entropy on the Mean (MEM) and in a Bayesian sense are reviewed, with some improvements on previous results. Then the MEM and Bayesian approaches are extended to default models with a dependence structure, yielding new families of solutions. One family involves a transfer kernel, and allows using prior information such as modality, convexity, or Sobolev norms. Another family of solutions with possibly nonconvex criteria, is arrived at using default models with exchangeable random variables. The main technical tools are convex analysis and large deviations theory.

Original languageEnglish
Pages (from-to)2253-2270
Number of pages18
JournalIEEE Transactions on Information Theory
Volume45
Issue number7
DOIs
Publication statusPublished - 1999

Fingerprint

entropy
Interpolation
Entropy
Pixels
Random variables
interpretation

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Information Systems

Cite this

MEM pixel correlated solutions for generalized moment and interpolation problems. / Csiszár, I.; Gamboa, Fabrice; Gassiat, Elisabeth.

In: IEEE Transactions on Information Theory, Vol. 45, No. 7, 1999, p. 2253-2270.

Research output: Contribution to journalArticle

Csiszár, I. ; Gamboa, Fabrice ; Gassiat, Elisabeth. / MEM pixel correlated solutions for generalized moment and interpolation problems. In: IEEE Transactions on Information Theory. 1999 ; Vol. 45, No. 7. pp. 2253-2270.
@article{f4530ec4ba6b455cb9487fd375b98d9c,
title = "MEM pixel correlated solutions for generalized moment and interpolation problems",
abstract = "In generalized moment problems (signed) measures are searched to fit given observations, or continuous functions are searched to fit given constraints. Known convex methods for solving such problems, and their stochastic interpretations via Maximum Entropy on the Mean (MEM) and in a Bayesian sense are reviewed, with some improvements on previous results. Then the MEM and Bayesian approaches are extended to default models with a dependence structure, yielding new families of solutions. One family involves a transfer kernel, and allows using prior information such as modality, convexity, or Sobolev norms. Another family of solutions with possibly nonconvex criteria, is arrived at using default models with exchangeable random variables. The main technical tools are convex analysis and large deviations theory.",
author = "I. Csisz{\'a}r and Fabrice Gamboa and Elisabeth Gassiat",
year = "1999",
doi = "10.1109/18.796367",
language = "English",
volume = "45",
pages = "2253--2270",
journal = "IEEE Transactions on Information Theory",
issn = "0018-9448",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "7",

}

TY - JOUR

T1 - MEM pixel correlated solutions for generalized moment and interpolation problems

AU - Csiszár, I.

AU - Gamboa, Fabrice

AU - Gassiat, Elisabeth

PY - 1999

Y1 - 1999

N2 - In generalized moment problems (signed) measures are searched to fit given observations, or continuous functions are searched to fit given constraints. Known convex methods for solving such problems, and their stochastic interpretations via Maximum Entropy on the Mean (MEM) and in a Bayesian sense are reviewed, with some improvements on previous results. Then the MEM and Bayesian approaches are extended to default models with a dependence structure, yielding new families of solutions. One family involves a transfer kernel, and allows using prior information such as modality, convexity, or Sobolev norms. Another family of solutions with possibly nonconvex criteria, is arrived at using default models with exchangeable random variables. The main technical tools are convex analysis and large deviations theory.

AB - In generalized moment problems (signed) measures are searched to fit given observations, or continuous functions are searched to fit given constraints. Known convex methods for solving such problems, and their stochastic interpretations via Maximum Entropy on the Mean (MEM) and in a Bayesian sense are reviewed, with some improvements on previous results. Then the MEM and Bayesian approaches are extended to default models with a dependence structure, yielding new families of solutions. One family involves a transfer kernel, and allows using prior information such as modality, convexity, or Sobolev norms. Another family of solutions with possibly nonconvex criteria, is arrived at using default models with exchangeable random variables. The main technical tools are convex analysis and large deviations theory.

UR - http://www.scopus.com/inward/record.url?scp=0033336975&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0033336975&partnerID=8YFLogxK

U2 - 10.1109/18.796367

DO - 10.1109/18.796367

M3 - Article

VL - 45

SP - 2253

EP - 2270

JO - IEEE Transactions on Information Theory

JF - IEEE Transactions on Information Theory

SN - 0018-9448

IS - 7

ER -