Mixtures of experts estimate a posteriori probabilities

Perry Moerland

doi:10.1007/bfb0020204

Mixtures of experts estimate a posteriori probabilities

Perry Moerland^*

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

Abstract

The mixtures of experts (ME) model offers a modular structure suitable for a divide-and-conquer approach to pattern recognition. It has a probabilistic interpretation in terms of a mixture model, which forms the basis for the error function associated with MEs. In this paper, it is shown that for classification problems the minimization of this ME error function leads to ME outputs estimating the a posteriori probabilities of class membership of the input vector.

Original language	English
Title of host publication	Artificial Neural Networks - ICANN 1997 - 7th International Conference, Proceeedings
Publisher	Springer Verlag
Pages	499-504
Number of pages	6
ISBN (Print)	3540636315, 9783540636311
DOIs	https://doi.org/10.1007/bfb0020204
Publication status	Published - 1997
Event	7th International Conference on Artificial Neural Networks, ICANN 1997 - Lausanne, Switzerland Duration: 8 Oct 1997 → 10 Oct 1997

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	1327
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	7th International Conference on Artificial Neural Networks, ICANN 1997
Country/Territory	Switzerland
City	Lausanne
Period	08/10/1997 → 10/10/1997

Access to Document

10.1007/bfb0020204

Cite this

Moerland, P. (1997). Mixtures of experts estimate a posteriori probabilities. In Artificial Neural Networks - ICANN 1997 - 7th International Conference, Proceeedings (pp. 499-504). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1327). Springer Verlag. https://doi.org/10.1007/bfb0020204

@inproceedings{0679e460769d4fbda1973f84627e6e69,

title = "Mixtures of experts estimate a posteriori probabilities",

abstract = "The mixtures of experts (ME) model offers a modular structure suitable for a divide-and-conquer approach to pattern recognition. It has a probabilistic interpretation in terms of a mixture model, which forms the basis for the error function associated with MEs. In this paper, it is shown that for classification problems the minimization of this ME error function leads to ME outputs estimating the a posteriori probabilities of class membership of the input vector.",

author = "Perry Moerland",

year = "1997",

doi = "10.1007/bfb0020204",

language = "English",

isbn = "3540636315",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "499--504",

booktitle = "Artificial Neural Networks - ICANN 1997 - 7th International Conference, Proceeedings",

address = "Germany",

note = "7th International Conference on Artificial Neural Networks, ICANN 1997 ; Conference date: 08-10-1997 Through 10-10-1997",

}

Moerland, P 1997, Mixtures of experts estimate a posteriori probabilities. in Artificial Neural Networks - ICANN 1997 - 7th International Conference, Proceeedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 1327, Springer Verlag, pp. 499-504, 7th International Conference on Artificial Neural Networks, ICANN 1997, Lausanne, Switzerland, 08/10/1997. https://doi.org/10.1007/bfb0020204

Mixtures of experts estimate a posteriori probabilities. / Moerland, Perry.
Artificial Neural Networks - ICANN 1997 - 7th International Conference, Proceeedings. Springer Verlag, 1997. p. 499-504 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1327).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Mixtures of experts estimate a posteriori probabilities

AU - Moerland, Perry

PY - 1997

Y1 - 1997

N2 - The mixtures of experts (ME) model offers a modular structure suitable for a divide-and-conquer approach to pattern recognition. It has a probabilistic interpretation in terms of a mixture model, which forms the basis for the error function associated with MEs. In this paper, it is shown that for classification problems the minimization of this ME error function leads to ME outputs estimating the a posteriori probabilities of class membership of the input vector.

AB - The mixtures of experts (ME) model offers a modular structure suitable for a divide-and-conquer approach to pattern recognition. It has a probabilistic interpretation in terms of a mixture model, which forms the basis for the error function associated with MEs. In this paper, it is shown that for classification problems the minimization of this ME error function leads to ME outputs estimating the a posteriori probabilities of class membership of the input vector.

UR - https://www.scopus.com/pages/publications/84956690779

U2 - 10.1007/bfb0020204

DO - 10.1007/bfb0020204

M3 - Conference contribution

SN - 3540636315

SN - 9783540636311

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 499

EP - 504

BT - Artificial Neural Networks - ICANN 1997 - 7th International Conference, Proceeedings

PB - Springer Verlag

T2 - 7th International Conference on Artificial Neural Networks, ICANN 1997

Y2 - 8 October 1997 through 10 October 1997

ER -

Mixtures of experts estimate a posteriori probabilities

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

Cite this