No free lunch for cross validation

Huaiyu Zhu; Richard Rohwer

doi:10.1162/neco.1996.8.7.14211

No free lunch for cross validation

Huaiyu Zhu, Richard Rohwer

Research output: Contribution to journal › Article › peer-review

Abstract

It is known theoretically that an algorithm cannot be good for an arbitrary prior. We show that in practical terms this also applies to the technique of ``cross validation'', which has been widely regarded as defying this general rule. Numerical examples are analysed in detail. Their implications to researches on learning algorithms are discussed.

Original language	English
Pages (from-to)	1421-1426
Number of pages	6
Journal	Neural Computation
Volume	8
Issue number	7
DOIs	https://doi.org/10.1162/neco.1996.8.7.14211
Publication status	Published - 1 Oct 1996

Bibliographical note

Copyright of the Massachusetts Institute of Technology Press (MIT Press)

Keywords

learning algorithms
cross validation

Access to Document

10.1162/neco.1996.8.7.14211

Cite this

@article{c9e25ef5b19243388ad5b6fa481eca0d,

title = "No free lunch for cross validation",

abstract = "It is known theoretically that an algorithm cannot be good for an arbitrary prior. We show that in practical terms this also applies to the technique of ``cross validation'', which has been widely regarded as defying this general rule. Numerical examples are analysed in detail. Their implications to researches on learning algorithms are discussed.",

keywords = "learning algorithms, cross validation",

author = "Huaiyu Zhu and Richard Rohwer",

note = "Copyright of the Massachusetts Institute of Technology Press (MIT Press)",

year = "1996",

month = oct,

day = "1",

doi = "10.1162/neco.1996.8.7.14211",

language = "English",

volume = "8",

pages = "1421--1426",

journal = "Neural Computation",

issn = "0899-7667",

publisher = "MIT Press Journals",

number = "7",

}

TY - JOUR

T1 - No free lunch for cross validation

AU - Zhu, Huaiyu

AU - Rohwer, Richard

N1 - Copyright of the Massachusetts Institute of Technology Press (MIT Press)

PY - 1996/10/1

Y1 - 1996/10/1

N2 - It is known theoretically that an algorithm cannot be good for an arbitrary prior. We show that in practical terms this also applies to the technique of ``cross validation'', which has been widely regarded as defying this general rule. Numerical examples are analysed in detail. Their implications to researches on learning algorithms are discussed.

AB - It is known theoretically that an algorithm cannot be good for an arbitrary prior. We show that in practical terms this also applies to the technique of ``cross validation'', which has been widely regarded as defying this general rule. Numerical examples are analysed in detail. Their implications to researches on learning algorithms are discussed.

KW - learning algorithms

KW - cross validation

UR - http://www.mitpressjournals.org/doi/abs/10.1162/neco.1996.8.7.1421?journalCode=neco

U2 - 10.1162/neco.1996.8.7.14211

DO - 10.1162/neco.1996.8.7.14211

M3 - Article

SN - 0899-7667

VL - 8

SP - 1421

EP - 1426

JO - Neural Computation

JF - Neural Computation

IS - 7

ER -

No free lunch for cross validation

Abstract

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this