MML invariant linear regression

Daniel F. Schmidt; Enes Makalic

doi:10.1007/978-3-642-10439-8_32

MML invariant linear regression

Daniel F. Schmidt, Enes Makalic

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

14 Citations (Scopus)

Abstract

This paper derives two new information theoretic linear regression criteria based on the minimum message length principle. Both criteria are invariant to full rank affine transformations of the design matrix and yield estimates that are minimax with respect to squared error loss. The new criteria are compared against state of the art information theoretic model selection criteria on both real and synthetic data and show good performance in all cases.

Original language	English
Title of host publication	AI 2009
Subtitle of host publication	Advances in Artificial Intelligence - 22nd Australasian Joint Conference, Proceedings
Publisher	Springer
Pages	312-321
Number of pages	10
ISBN (Print)	364210438X, 9783642104381
DOIs	https://doi.org/10.1007/978-3-642-10439-8_32
Publication status	Published - 2009
Externally published	Yes
Event	Australasian Joint Conference on Artificial Intelligence 2009 - Melbourne, Australia Duration: 1 Dec 2009 → 4 Dec 2009 Conference number: 22nd https://link.springer.com/book/10.1007/978-3-642-10439-8 (Proceedings)

Publication series

Name	Lecture Notes in Computer Science
Publisher	Springer
Volume	5866
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	Australasian Joint Conference on Artificial Intelligence 2009
Abbreviated title	AI 2009
Country/Territory	Australia
City	Melbourne
Period	1/12/09 → 4/12/09
Internet address	https://link.springer.com/book/10.1007/978-3-642-10439-8 (Proceedings)

Access to Document

10.1007/978-3-642-10439-8_32

Cite this

@inproceedings{91648ed575044a029bdb8396b3e30376,

title = "MML invariant linear regression",

abstract = "This paper derives two new information theoretic linear regression criteria based on the minimum message length principle. Both criteria are invariant to full rank affine transformations of the design matrix and yield estimates that are minimax with respect to squared error loss. The new criteria are compared against state of the art information theoretic model selection criteria on both real and synthetic data and show good performance in all cases.",

author = "Schmidt, \{Daniel F.\} and Enes Makalic",

year = "2009",

doi = "10.1007/978-3-642-10439-8\_32",

language = "English",

isbn = "364210438X",

series = "Lecture Notes in Computer Science",

publisher = "Springer",

pages = "312--321",

booktitle = "AI 2009",

address = "Switzerland",

note = "Australasian Joint Conference on Artificial Intelligence 2009, AI 2009 ; Conference date: 01-12-2009 Through 04-12-2009",

url = "https://link.springer.com/book/10.1007/978-3-642-10439-8",

}

Schmidt, DF & Makalic, E 2009, MML invariant linear regression. in AI 2009: Advances in Artificial Intelligence - 22nd Australasian Joint Conference, Proceedings. Lecture Notes in Computer Science, vol. 5866 , Springer, pp. 312-321, Australasian Joint Conference on Artificial Intelligence 2009, Melbourne, Victoria, Australia, 1/12/09. https://doi.org/10.1007/978-3-642-10439-8_32

TY - GEN

T1 - MML invariant linear regression

AU - Schmidt, Daniel F.

AU - Makalic, Enes

N1 - Conference code: 22nd

PY - 2009

Y1 - 2009

N2 - This paper derives two new information theoretic linear regression criteria based on the minimum message length principle. Both criteria are invariant to full rank affine transformations of the design matrix and yield estimates that are minimax with respect to squared error loss. The new criteria are compared against state of the art information theoretic model selection criteria on both real and synthetic data and show good performance in all cases.

AB - This paper derives two new information theoretic linear regression criteria based on the minimum message length principle. Both criteria are invariant to full rank affine transformations of the design matrix and yield estimates that are minimax with respect to squared error loss. The new criteria are compared against state of the art information theoretic model selection criteria on both real and synthetic data and show good performance in all cases.

UR - http://www.scopus.com/inward/record.url?scp=78650423943&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-10439-8_32

DO - 10.1007/978-3-642-10439-8_32

M3 - Conference Paper

AN - SCOPUS:78650423943

SN - 364210438X

SN - 9783642104381

T3 - Lecture Notes in Computer Science

SP - 312

EP - 321

BT - AI 2009

PB - Springer

T2 - Australasian Joint Conference on Artificial Intelligence 2009

Y2 - 1 December 2009 through 4 December 2009

ER -

MML invariant linear regression

Abstract

Publication series

Conference

Access to Document

Other files and links

Cite this