Early Stop Criterion from the Bootstrap Ensemble

Lars Kai Hansen; Jan Larsen; Torben L. Fog

doi:10.1109/ICASSP.1997.595474

Early Stop Criterion from the Bootstrap Ensemble

Lars Kai Hansen, Jan Larsen, Torben L. Fog

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

561 Downloads (Orbit)

Abstract

This paper addresses the problem of generalization error estimation in neural networks. A new early stop criterion based on a Bootstrap estimate of the generalization error is suggested. The estimate does not require the network to be trained to the minimum of the cost function, as required by other methods based on asymptotic theory. Moreover, in contrast to methods based on cross-validation which require data left out for testing, and thus biasing the estimate, the Bootstrap technique does not have this disadvantage. The potential of the suggested technique is demonstrated on various time-series problems

Original language	English
Title of host publication	Proceedings of IEEE ICASSP'97
Place of Publication	Munich, Germany
Publisher	IEEE
Publication date	1997
Pages	3205-3208
ISBN (Print)	0-8186-7919-0
DOIs	https://doi.org/10.1109/ICASSP.1997.595474
Publication status	Published - 1997
Event	1997 IEEE International Conference on Acoustics, Speech, and Signal Processing - Munich, Germany Duration: 21 Apr 1997 → 24 Apr 1997 https://ieeexplore.ieee.org/xpl/conhome/4635/proceeding

Conference

Conference	1997 IEEE International Conference on Acoustics, Speech, and Signal Processing
Country/Territory	Germany
City	Munich
Period	21/04/1997 → 24/04/1997
Internet address	https://ieeexplore.ieee.org/xpl/conhome/4635/proceeding

Bibliographical note

Copyright: 1997 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE

Access to Document

10.1109/ICASSP.1997.595474

HansenFinal published version, 363 KB

OpenUrl availability

Full text

Cite this

@inproceedings{48e21e8fa4f84797956834d987ef86a4,

title = "Early Stop Criterion from the Bootstrap Ensemble",

abstract = "This paper addresses the problem of generalization error estimation in neural networks. A new early stop criterion based on a Bootstrap estimate of the generalization error is suggested. The estimate does not require the network to be trained to the minimum of the cost function, as required by other methods based on asymptotic theory. Moreover, in contrast to methods based on cross-validation which require data left out for testing, and thus biasing the estimate, the Bootstrap technique does not have this disadvantage. The potential of the suggested technique is demonstrated on various time-series problems",

author = "Hansen, {Lars Kai} and Jan Larsen and Fog, {Torben L.}",

note = "Copyright: 1997 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE; 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing, IEEE ICASSP'97 ; Conference date: 21-04-1997 Through 24-04-1997",

year = "1997",

doi = "10.1109/ICASSP.1997.595474",

language = "English",

isbn = "0-8186-7919-0",

pages = "3205--3208",

booktitle = "Proceedings of IEEE ICASSP'97",

publisher = "IEEE",

address = "United States",

url = "https://ieeexplore.ieee.org/xpl/conhome/4635/proceeding",

}

TY - GEN

T1 - Early Stop Criterion from the Bootstrap Ensemble

AU - Hansen, Lars Kai

AU - Larsen, Jan

AU - Fog, Torben L.

N1 - Copyright: 1997 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE

PY - 1997

Y1 - 1997

N2 - This paper addresses the problem of generalization error estimation in neural networks. A new early stop criterion based on a Bootstrap estimate of the generalization error is suggested. The estimate does not require the network to be trained to the minimum of the cost function, as required by other methods based on asymptotic theory. Moreover, in contrast to methods based on cross-validation which require data left out for testing, and thus biasing the estimate, the Bootstrap technique does not have this disadvantage. The potential of the suggested technique is demonstrated on various time-series problems

AB - This paper addresses the problem of generalization error estimation in neural networks. A new early stop criterion based on a Bootstrap estimate of the generalization error is suggested. The estimate does not require the network to be trained to the minimum of the cost function, as required by other methods based on asymptotic theory. Moreover, in contrast to methods based on cross-validation which require data left out for testing, and thus biasing the estimate, the Bootstrap technique does not have this disadvantage. The potential of the suggested technique is demonstrated on various time-series problems

U2 - 10.1109/ICASSP.1997.595474

DO - 10.1109/ICASSP.1997.595474

M3 - Article in proceedings

SN - 0-8186-7919-0

SP - 3205

EP - 3208

BT - Proceedings of IEEE ICASSP'97

PB - IEEE

CY - Munich, Germany

T2 - 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing

Y2 - 21 April 1997 through 24 April 1997

ER -

Early Stop Criterion from the Bootstrap Ensemble

Abstract

Conference

Bibliographical note

Access to Document

OpenUrl availability

Fingerprint

Cite this