A reward-directed Bayesian classifier

Li Hui; Liao Xuejun; Lawrence Carin

A reward-directed Bayesian classifier

Li Hui, Liao Xuejun, Lawrence Carin

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

We consider a classification problem wherein the class features are not given a priori. The classifier is responsible for selecting the features, to minimize the cost of observing features while also maximizing the classification performance. We propose a reward-directed Bayesian classifier (RDBC) to solve this problem. The RDBC features an internal state structure for preserving the feature dependence, and is formulated as a partially observable Markov decision process (POMDP). The results on a diabetes dataset show the RDBC with a moderate number of states significantly improves over the naive Bayes classifier, both in prediction accuracy and observation parsimony. It is also demonstrated that the RDBC performs better by using more states to increase its memory. © 2006 IEEE.

Original language	English (US)
Title of host publication	ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
State	Published - Dec 1 2006
Externally published	Yes

Bibliographical note

Generated from Scopus record by KAUST IRTS on 2021-02-09

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Cite this

@inproceedings{5fae46a9c3c04b1b90d2afadfbf2af03,

title = "A reward-directed Bayesian classifier",

abstract = "We consider a classification problem wherein the class features are not given a priori. The classifier is responsible for selecting the features, to minimize the cost of observing features while also maximizing the classification performance. We propose a reward-directed Bayesian classifier (RDBC) to solve this problem. The RDBC features an internal state structure for preserving the feature dependence, and is formulated as a partially observable Markov decision process (POMDP). The results on a diabetes dataset show the RDBC with a moderate number of states significantly improves over the naive Bayes classifier, both in prediction accuracy and observation parsimony. It is also demonstrated that the RDBC performs better by using more states to increase its memory. {\textcopyright} 2006 IEEE.",

author = "Li Hui and Liao Xuejun and Lawrence Carin",

note = "Generated from Scopus record by KAUST IRTS on 2021-02-09",

year = "2006",

month = dec,

day = "1",

language = "English (US)",

isbn = "142440469X",

booktitle = "ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings",

}

TY - GEN

T1 - A reward-directed Bayesian classifier

AU - Hui, Li

AU - Xuejun, Liao

AU - Carin, Lawrence

N1 - Generated from Scopus record by KAUST IRTS on 2021-02-09

PY - 2006/12/1

Y1 - 2006/12/1

N2 - We consider a classification problem wherein the class features are not given a priori. The classifier is responsible for selecting the features, to minimize the cost of observing features while also maximizing the classification performance. We propose a reward-directed Bayesian classifier (RDBC) to solve this problem. The RDBC features an internal state structure for preserving the feature dependence, and is formulated as a partially observable Markov decision process (POMDP). The results on a diabetes dataset show the RDBC with a moderate number of states significantly improves over the naive Bayes classifier, both in prediction accuracy and observation parsimony. It is also demonstrated that the RDBC performs better by using more states to increase its memory. © 2006 IEEE.

AB - We consider a classification problem wherein the class features are not given a priori. The classifier is responsible for selecting the features, to minimize the cost of observing features while also maximizing the classification performance. We propose a reward-directed Bayesian classifier (RDBC) to solve this problem. The RDBC features an internal state structure for preserving the feature dependence, and is formulated as a partially observable Markov decision process (POMDP). The results on a diabetes dataset show the RDBC with a moderate number of states significantly improves over the naive Bayes classifier, both in prediction accuracy and observation parsimony. It is also demonstrated that the RDBC performs better by using more states to increase its memory. © 2006 IEEE.

UR - http://www.scopus.com/inward/record.url?scp=33947663057&partnerID=8YFLogxK

M3 - Conference contribution

SN - 142440469X

BT - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

ER -

A reward-directed Bayesian classifier

Abstract

Bibliographical note

UN SDGs

Other files and links

Fingerprint

Cite this