Exploiting informative priors for Bayesian classification and regression trees

N. Angelopoulos; J. Cussens; L. Pack Kaelbling; A. Saffiotti

Exploiting informative priors for Bayesian classification and regression trees

N. Angelopoulos, J. Cussens, L. Pack Kaelbling (Editor), A. Saffiotti (Editor)

Computer Science

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

Abstract

A general method for defining informative priors on statistical models is presented and applied specifically to the space of classification and regression trees. A Bayesian approach to learning such models from data is taken, with the Metropolis- Hastings algorithm being used to approximately sample from the posterior. By only using proposal distributions closely tied to the prior, acceptance probabilities are easily computable via marginal likelihood ratios, whatever the prior used. Our approach is empirically tested by varying (i) the data, (ii) the prior and (iii) the proposal distribution. A comparison with related work is given.

Original language	English
Title of host publication	Proceedings of the Nineteenth International Joint Conference on Artificial Intelligence
Publisher	Professional Book Center
Pages	641-646
Number of pages	5
ISBN (Print)	0938075934
Publication status	Published - 2005
Event	IJCAI-05 - Edinburgh, Scotland Duration: 5 Jul 2005 → …

Conference

Conference	IJCAI-05
City	Edinburgh, Scotland
Period	5/07/05 → …

Access to Document

http://www.ijcai.org/papers/1013.pdf

Cite this

@inproceedings{39f7d53429c248809dfb5b314861b330,

title = "Exploiting informative priors for Bayesian classification and regression trees",

abstract = "A general method for defining informative priors on statistical models is presented and applied specifically to the space of classification and regression trees. A Bayesian approach to learning such models from data is taken, with the Metropolis- Hastings algorithm being used to approximately sample from the posterior. By only using proposal distributions closely tied to the prior, acceptance probabilities are easily computable via marginal likelihood ratios, whatever the prior used. Our approach is empirically tested by varying (i) the data, (ii) the prior and (iii) the proposal distribution. A comparison with related work is given.",

author = "N. Angelopoulos and J. Cussens and {Pack Kaelbling}, L. and A. Saffiotti",

year = "2005",

language = "English",

isbn = "0938075934",

pages = "641--646",

booktitle = "Proceedings of the Nineteenth International Joint Conference on Artificial Intelligence",

publisher = "Professional Book Center",

note = "IJCAI-05 ; Conference date: 05-07-2005",

}

TY - GEN

T1 - Exploiting informative priors for Bayesian classification and regression trees

AU - Angelopoulos, N.

AU - Cussens, J.

A2 - Pack Kaelbling, L.

A2 - Saffiotti, A.

PY - 2005

Y1 - 2005

N2 - A general method for defining informative priors on statistical models is presented and applied specifically to the space of classification and regression trees. A Bayesian approach to learning such models from data is taken, with the Metropolis- Hastings algorithm being used to approximately sample from the posterior. By only using proposal distributions closely tied to the prior, acceptance probabilities are easily computable via marginal likelihood ratios, whatever the prior used. Our approach is empirically tested by varying (i) the data, (ii) the prior and (iii) the proposal distribution. A comparison with related work is given.

AB - A general method for defining informative priors on statistical models is presented and applied specifically to the space of classification and regression trees. A Bayesian approach to learning such models from data is taken, with the Metropolis- Hastings algorithm being used to approximately sample from the posterior. By only using proposal distributions closely tied to the prior, acceptance probabilities are easily computable via marginal likelihood ratios, whatever the prior used. Our approach is empirically tested by varying (i) the data, (ii) the prior and (iii) the proposal distribution. A comparison with related work is given.

M3 - Conference contribution

SN - 0938075934

SP - 641

EP - 646

BT - Proceedings of the Nineteenth International Joint Conference on Artificial Intelligence

PB - Professional Book Center

T2 - IJCAI-05

Y2 - 5 July 2005

ER -