Prediction and modeling with partial dependencies

T.J. Tjalkens

doi:10.1109/ITA.2008.4601082

Prediction and modeling with partial dependencies

T.J. Tjalkens

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

1 Citation (Scopus)

86 Downloads (Pure)

Abstract

The author consider a binary classification problem with a feature vector of high dimensionality. Spam mail filters are a popular example hereof. A Bayesian approach requires us to estimate the probability of a feature vector given the class of the object. Due to the size of the feature vector this is an unfeasible task. A useful approach is to split the feature space into several (conditionally) independent subspaces. This results in a new problem, namely how to find the ldquobestrdquo subdivision. In this paper the author consider a weighing approach that will perform (asymptotically) as good as the best subdivision and still has a manageable complexity.

Original language	English
Title of host publication	Information Theory and Applications Workshop, 2008 , 3rd ,27 January -1 February 2008, San Diego, U.S.A.
Place of Publication	Piscataway
Publisher	Institute of Electrical and Electronics Engineers
Pages	1-11
ISBN (Print)	978-1-4244-2670-6
DOIs	https://doi.org/10.1109/ITA.2008.4601082
Publication status	Published - 2008

Access to Document

10.1109/ITA.2008.4601082

Metis223105Final published version, 1.32 MB

Cite this

@inproceedings{e524836e3cc24d5d88be5e03180c7106,

title = "Prediction and modeling with partial dependencies",

abstract = "The author consider a binary classification problem with a feature vector of high dimensionality. Spam mail filters are a popular example hereof. A Bayesian approach requires us to estimate the probability of a feature vector given the class of the object. Due to the size of the feature vector this is an unfeasible task. A useful approach is to split the feature space into several (conditionally) independent subspaces. This results in a new problem, namely how to find the ldquobestrdquo subdivision. In this paper the author consider a weighing approach that will perform (asymptotically) as good as the best subdivision and still has a manageable complexity.",

author = "T.J. Tjalkens",

year = "2008",

doi = "10.1109/ITA.2008.4601082",

language = "English",

isbn = "978-1-4244-2670-6",

pages = "1--11",

booktitle = "Information Theory and Applications Workshop, 2008 , 3rd ,27 January -1 February 2008, San Diego, U.S.A.",

publisher = "Institute of Electrical and Electronics Engineers",

address = "United States",

}

Prediction and modeling with partial dependencies. / Tjalkens, T.J.

Information Theory and Applications Workshop, 2008 , 3rd ,27 January -1 February 2008, San Diego, U.S.A.. Piscataway : Institute of Electrical and Electronics Engineers, 2008. p. 1-11.

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

TY - GEN

T1 - Prediction and modeling with partial dependencies

AU - Tjalkens, T.J.

PY - 2008

Y1 - 2008

N2 - The author consider a binary classification problem with a feature vector of high dimensionality. Spam mail filters are a popular example hereof. A Bayesian approach requires us to estimate the probability of a feature vector given the class of the object. Due to the size of the feature vector this is an unfeasible task. A useful approach is to split the feature space into several (conditionally) independent subspaces. This results in a new problem, namely how to find the ldquobestrdquo subdivision. In this paper the author consider a weighing approach that will perform (asymptotically) as good as the best subdivision and still has a manageable complexity.

AB - The author consider a binary classification problem with a feature vector of high dimensionality. Spam mail filters are a popular example hereof. A Bayesian approach requires us to estimate the probability of a feature vector given the class of the object. Due to the size of the feature vector this is an unfeasible task. A useful approach is to split the feature space into several (conditionally) independent subspaces. This results in a new problem, namely how to find the ldquobestrdquo subdivision. In this paper the author consider a weighing approach that will perform (asymptotically) as good as the best subdivision and still has a manageable complexity.

U2 - 10.1109/ITA.2008.4601082

DO - 10.1109/ITA.2008.4601082

M3 - Conference contribution

SN - 978-1-4244-2670-6

SP - 1

EP - 11

BT - Information Theory and Applications Workshop, 2008 , 3rd ,27 January -1 February 2008, San Diego, U.S.A.

PB - Institute of Electrical and Electronics Engineers

CY - Piscataway

ER -

Prediction and modeling with partial dependencies

Abstract

Access to Document

Fingerprint

Cite this