Presentation 2014-06-25
A Discriminative Model resistant to Malicious Annotators
Atsutoshi KUMAGAI, Shingo ORIHARA, Yasushi OKANO, Tomoharu IWATA, Yoshihito OSHIMA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Recently, there have been a lot of studies on learning a classifier from a large amount of labeled data collected from crowds. However, the existing methods have a problem that the accuracy of the classifier drastically deteriorates if there are malicious annotators intentionally giving wrong labels. In this paper, to solve this problem, we propose a method of learning a classifier resistant to malicious annotators by introducing degrees of similarity between discriminative models of annotators. Through experiments, we show the effectiveness of our proposed method.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Machine Learning / multiple annotators / Empirical Bayes method
Paper # NC2014-1,IBISML2014-1
Date of Issue

Conference Information
Committee NC
Conference Date 2014/6/18(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A Discriminative Model resistant to Malicious Annotators
Sub Title (in English)
Keyword(1) Machine Learning
Keyword(2) multiple annotators
Keyword(3) Empirical Bayes method
1st Author's Name Atsutoshi KUMAGAI
1st Author's Affiliation NTT Secure Platform Laboratories, NTT Corporation()
2nd Author's Name Shingo ORIHARA
2nd Author's Affiliation NTT Secure Platform Laboratories, NTT Corporation
3rd Author's Name Yasushi OKANO
3rd Author's Affiliation NTT Secure Platform Laboratories, NTT Corporation
4th Author's Name Tomoharu IWATA
4th Author's Affiliation NTT Communication Science Laboratories, NTT Corporation
5th Author's Name Yoshihito OSHIMA
5th Author's Affiliation NTT Secure Platform Laboratories, NTT Corporation
Date 2014-06-25
Paper # NC2014-1,IBISML2014-1
Volume (vol) vol.114
Number (no) 104
Page pp.pp.-
#Pages 7
Date of Issue