MassBayes: a new generative classifier with multi-dimensional likelihood estimation

Sunil Aryal; Kai Ming Ting

doi:10.1007/978-3-642-37453-1

MassBayes: a new generative classifier with multi-dimensional likelihood estimation

Sunil Aryal, Kai Ming Ting

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

3 Citations (Scopus)

Abstract

Existing generative classifiers (e.g., BayesNet and AnDE) make independence assumptions and estimate one-dimensional likelihood. This paper presents a new generative classifier called MassBayes that estimates multi-dimensional likelihood without making any explicit assumptions. It aggregates the multi-dimensional likelihoods estimated from random subsets of the training data using varying size random feature subsets. Our empirical evaluations show that MassBayes yields better classification accuracy than the existing generative classifiers in large data sets. As it works with fixed-size subsets of training data, it has constant training time complexity and constant space complexity, and it can easily scale up to very large data sets.

Original language	English
Title of host publication	Advances in Knowledge Discovery and Data Mining: 17th Pacific-Asia Conference, PAKDD 2013 Proceedings, Part I
Editors	Jian Pei, Vincent S Tseng, Longbing Cao, Hiroshi Motoda, Guandong Xu
Place of Publication	Germany
Publisher	Springer
Pages	136 - 148
Number of pages	13
ISBN (Print)	9783642374524
DOIs	https://doi.org/10.1007/978-3-642-37453-1
Publication status	Published - 2013
Event	Pacific-Asia Conference on Knowledge Discovery and Data Mining 2013 - Gold Coast, Australia Duration: 14 Apr 2013 → 17 Apr 2013 Conference number: 17th https://link.springer.com/book/10.1007/978-3-642-37453-1 (Proceedings)

Conference

Conference	Pacific-Asia Conference on Knowledge Discovery and Data Mining 2013
Abbreviated title	PAKDD 2013
Country/Territory	Australia
City	Gold Coast
Period	14/04/13 → 17/04/13
Internet address	https://link.springer.com/book/10.1007/978-3-642-37453-1 (Proceedings)

Access to Document

10.1007/978-3-642-37453-1

3 Citations
1 Article

A Generic Ensemble Approach to Estimate Multidimensional Likelihood in Bayesian Classifier Learning
Aryal, S. & Ting, K. M., 1 Aug 2016, In: Computational Intelligence. 32, 3, p. 458-479 22 p.
Research output: Contribution to journal › Article › Research › peer-review

Cite this

@inproceedings{267cef253d06441c902ba9904f58c5dc,

title = "MassBayes: a new generative classifier with multi-dimensional likelihood estimation",

abstract = "Existing generative classifiers (e.g., BayesNet and AnDE) make independence assumptions and estimate one-dimensional likelihood. This paper presents a new generative classifier called MassBayes that estimates multi-dimensional likelihood without making any explicit assumptions. It aggregates the multi-dimensional likelihoods estimated from random subsets of the training data using varying size random feature subsets. Our empirical evaluations show that MassBayes yields better classification accuracy than the existing generative classifiers in large data sets. As it works with fixed-size subsets of training data, it has constant training time complexity and constant space complexity, and it can easily scale up to very large data sets.",

author = "Sunil Aryal and Ting, {Kai Ming}",

year = "2013",

doi = "10.1007/978-3-642-37453-1",

language = "English",

isbn = "9783642374524",

pages = "136 -- 148",

editor = "Jian Pei and Tseng, {Vincent S} and Longbing Cao and Hiroshi Motoda and Guandong Xu",

booktitle = "Advances in Knowledge Discovery and Data Mining: 17th Pacific-Asia Conference, PAKDD 2013 Proceedings, Part I",

publisher = "Springer",

note = "Pacific-Asia Conference on Knowledge Discovery and Data Mining 2013, PAKDD 2013 ; Conference date: 14-04-2013 Through 17-04-2013",

url = "https://link.springer.com/book/10.1007/978-3-642-37453-1",

}

Aryal, S & Ting, KM 2013, MassBayes: a new generative classifier with multi-dimensional likelihood estimation. in J Pei, VS Tseng, L Cao, H Motoda & G Xu (eds), Advances in Knowledge Discovery and Data Mining: 17th Pacific-Asia Conference, PAKDD 2013 Proceedings, Part I. Springer, Germany, pp. 136 - 148, Pacific-Asia Conference on Knowledge Discovery and Data Mining 2013, Gold Coast, Queensland, Australia, 14/04/13. https://doi.org/10.1007/978-3-642-37453-1

MassBayes: a new generative classifier with multi-dimensional likelihood estimation. / Aryal, Sunil; Ting, Kai Ming.
Advances in Knowledge Discovery and Data Mining: 17th Pacific-Asia Conference, PAKDD 2013 Proceedings, Part I. ed. / Jian Pei; Vincent S Tseng; Longbing Cao; Hiroshi Motoda; Guandong Xu. Germany: Springer, 2013. p. 136 - 148.

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

TY - GEN

T1 - MassBayes: a new generative classifier with multi-dimensional likelihood estimation

AU - Aryal, Sunil

AU - Ting, Kai Ming

N1 - Conference code: 17th

PY - 2013

Y1 - 2013

N2 - Existing generative classifiers (e.g., BayesNet and AnDE) make independence assumptions and estimate one-dimensional likelihood. This paper presents a new generative classifier called MassBayes that estimates multi-dimensional likelihood without making any explicit assumptions. It aggregates the multi-dimensional likelihoods estimated from random subsets of the training data using varying size random feature subsets. Our empirical evaluations show that MassBayes yields better classification accuracy than the existing generative classifiers in large data sets. As it works with fixed-size subsets of training data, it has constant training time complexity and constant space complexity, and it can easily scale up to very large data sets.

AB - Existing generative classifiers (e.g., BayesNet and AnDE) make independence assumptions and estimate one-dimensional likelihood. This paper presents a new generative classifier called MassBayes that estimates multi-dimensional likelihood without making any explicit assumptions. It aggregates the multi-dimensional likelihoods estimated from random subsets of the training data using varying size random feature subsets. Our empirical evaluations show that MassBayes yields better classification accuracy than the existing generative classifiers in large data sets. As it works with fixed-size subsets of training data, it has constant training time complexity and constant space complexity, and it can easily scale up to very large data sets.

U2 - 10.1007/978-3-642-37453-1

DO - 10.1007/978-3-642-37453-1

M3 - Conference Paper

SN - 9783642374524

SP - 136

EP - 148

BT - Advances in Knowledge Discovery and Data Mining: 17th Pacific-Asia Conference, PAKDD 2013 Proceedings, Part I

A2 - Pei, Jian

A2 - Tseng, Vincent S

A2 - Cao, Longbing

A2 - Motoda, Hiroshi

A2 - Xu, Guandong

PB - Springer

CY - Germany

T2 - Pacific-Asia Conference on Knowledge Discovery and Data Mining 2013

Y2 - 14 April 2013 through 17 April 2013

ER -

MassBayes: a new generative classifier with multi-dimensional likelihood estimation

Abstract

Conference

Access to Document

Research output

A Generic Ensemble Approach to Estimate Multidimensional Likelihood in Bayesian Classifier Learning

Cite this