A robust boosting method for mislabeled data

Natsuki Sano, Hideo Suzuki, Masato Koda

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

We propose a new, robust boosting method by using a siginoidal function as a loss function. In deriving the method, the stagewise additive modelling methodology is blended with the gradient descent algorithms. Based on intensive numerical experiments, we show that the proposed method is actually better than AdaBoost and other regularized method in test error rates in the case of noisy, mislabeled situation.

Original languageEnglish
Pages (from-to)182-196
Number of pages15
JournalJournal of the Operations Research Society of Japan
Volume47
Issue number3
Publication statusPublished - 2004 Sep
Externally publishedYes

Fingerprint

Boosting
Modeling methodology
Loss function
Numerical experiment
Gradient

Keywords

  • AdaBoost
  • Boosting
  • Data analysis
  • Data mining
  • Machine learning
  • Sigmoidal function

ASJC Scopus subject areas

  • Decision Sciences(all)
  • Management Science and Operations Research

Cite this

A robust boosting method for mislabeled data. / Sano, Natsuki; Suzuki, Hideo; Koda, Masato.

In: Journal of the Operations Research Society of Japan, Vol. 47, No. 3, 09.2004, p. 182-196.

Research output: Contribution to journalArticle

Sano, Natsuki ; Suzuki, Hideo ; Koda, Masato. / A robust boosting method for mislabeled data. In: Journal of the Operations Research Society of Japan. 2004 ; Vol. 47, No. 3. pp. 182-196.
@article{82347642dd6e4fba90e281ebf11e425c,
title = "A robust boosting method for mislabeled data",
abstract = "We propose a new, robust boosting method by using a siginoidal function as a loss function. In deriving the method, the stagewise additive modelling methodology is blended with the gradient descent algorithms. Based on intensive numerical experiments, we show that the proposed method is actually better than AdaBoost and other regularized method in test error rates in the case of noisy, mislabeled situation.",
keywords = "AdaBoost, Boosting, Data analysis, Data mining, Machine learning, Sigmoidal function",
author = "Natsuki Sano and Hideo Suzuki and Masato Koda",
year = "2004",
month = "9",
language = "English",
volume = "47",
pages = "182--196",
journal = "Journal of the Operations Research Society of Japan",
issn = "0453-4514",
publisher = "Operations Research Society of Japan",
number = "3",

}

TY - JOUR

T1 - A robust boosting method for mislabeled data

AU - Sano, Natsuki

AU - Suzuki, Hideo

AU - Koda, Masato

PY - 2004/9

Y1 - 2004/9

N2 - We propose a new, robust boosting method by using a siginoidal function as a loss function. In deriving the method, the stagewise additive modelling methodology is blended with the gradient descent algorithms. Based on intensive numerical experiments, we show that the proposed method is actually better than AdaBoost and other regularized method in test error rates in the case of noisy, mislabeled situation.

AB - We propose a new, robust boosting method by using a siginoidal function as a loss function. In deriving the method, the stagewise additive modelling methodology is blended with the gradient descent algorithms. Based on intensive numerical experiments, we show that the proposed method is actually better than AdaBoost and other regularized method in test error rates in the case of noisy, mislabeled situation.

KW - AdaBoost

KW - Boosting

KW - Data analysis

KW - Data mining

KW - Machine learning

KW - Sigmoidal function

UR - http://www.scopus.com/inward/record.url?scp=7244257643&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=7244257643&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:7244257643

VL - 47

SP - 182

EP - 196

JO - Journal of the Operations Research Society of Japan

JF - Journal of the Operations Research Society of Japan

SN - 0453-4514

IS - 3

ER -