Optimal Cross-Validation Split Ratio: Experimental
                Investigation

Cyril Goutte; Jan Larsen

Optimal Cross-Validation Split Ratio: Experimental Investigation

Cyril Goutte, Jan Larsen

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

Abstract

Cross-validation is a common method for assessing the generalisation ability of a model in order to tune a regularisation parameter or otherhyper-parameters of a learning process. The use of cross-validation requires to set yet an additional parameter, the split rati. While a few texts haveinvestigated theoretically the asymptotic setting of this ratio, no consensus has emerged. In this contribution, we investigate the sensitivity and optimalsetting of the split ratio on a particular model, a non-parametric kernel estimator with adaptive metric.

Original language	English
Title of host publication	Proceedings of ICANN´98
Place of Publication	London
Publisher	Springer
Publication date	1998
Pages	681-686
Publication status	Published - 1998
Event	ICANN´98, Proceedings of the 8th Int.Conf. on Artificial Neural Networks - Skoevde, Sweden Duration: 1 Jan 1998 → …

Conference

Conference	ICANN´98, Proceedings of the 8th Int.Conf. on Artificial Neural Networks
City	Skoevde, Sweden
Period	01/01/1998 → …

Cite this

@inproceedings{ebd98701bd684f64a9f62854e5259717,

title = "Optimal Cross-Validation Split Ratio: Experimental Investigation",

abstract = "Cross-validation is a common method for assessing the generalisation ability of a model in order to tune a regularisation parameter or otherhyper-parameters of a learning process. The use of cross-validation requires to set yet an additional parameter, the split rati. While a few texts haveinvestigated theoretically the asymptotic setting of this ratio, no consensus has emerged. In this contribution, we investigate the sensitivity and optimalsetting of the split ratio on a particular model, a non-parametric kernel estimator with adaptive metric.",

author = "Cyril Goutte and Jan Larsen",

year = "1998",

language = "English",

pages = "681--686",

booktitle = "Proceedings of ICANN´98",

publisher = "Springer",

note = "ICANN´98, Proceedings of the 8th Int.Conf. on Artificial Neural Networks ; Conference date: 01-01-1998",

}

TY - GEN

T1 - Optimal Cross-Validation Split Ratio: Experimental Investigation

AU - Goutte, Cyril

AU - Larsen, Jan

PY - 1998

Y1 - 1998

N2 - Cross-validation is a common method for assessing the generalisation ability of a model in order to tune a regularisation parameter or otherhyper-parameters of a learning process. The use of cross-validation requires to set yet an additional parameter, the split rati. While a few texts haveinvestigated theoretically the asymptotic setting of this ratio, no consensus has emerged. In this contribution, we investigate the sensitivity and optimalsetting of the split ratio on a particular model, a non-parametric kernel estimator with adaptive metric.

AB - Cross-validation is a common method for assessing the generalisation ability of a model in order to tune a regularisation parameter or otherhyper-parameters of a learning process. The use of cross-validation requires to set yet an additional parameter, the split rati. While a few texts haveinvestigated theoretically the asymptotic setting of this ratio, no consensus has emerged. In this contribution, we investigate the sensitivity and optimalsetting of the split ratio on a particular model, a non-parametric kernel estimator with adaptive metric.

M3 - Article in proceedings

SP - 681

EP - 686

BT - Proceedings of ICANN´98

PB - Springer

CY - London

T2 - ICANN´98, Proceedings of the 8th Int.Conf. on Artificial Neural Networks

Y2 - 1 January 1998

ER -