Presentation 2015-01-29
Metromaps as a Tool for Minimizing Human Interaction with Learning Bayesian Classifiers
Marat ZHANIKEEV,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) The traditional use of Bayesian classifiers is to have one period of learning followed by continuous application of the classifier to a much wider set of documents. However, continuous use of classifiers is also common when context grows over long periods of time. In the latter case, it is common to use classifier output as a recommendation which humans can accept as is or change when the classification is wrong. This paper proposes an interface that minimizes human interaction in such environments. The core concept is based on metromaps where documents are assigned to "train lines" first while the overall certainty of classification is based on documents' affiliation with multiple lines.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) metromaps / Bayesian classification / multidimensional classification / learning automation, folksonomy / ontologies / semantic web
Paper # NC2014-59
Date of Issue

Conference Information
Committee NC
Conference Date 2015/1/22(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 ENG
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Metromaps as a Tool for Minimizing Human Interaction with Learning Bayesian Classifiers
Sub Title (in English)
Keyword(1) metromaps
Keyword(2) Bayesian classification
Keyword(3) multidimensional classification
Keyword(4) learning automation, folksonomy
Keyword(5) ontologies
Keyword(6) semantic web
1st Author's Name Marat ZHANIKEEV
1st Author's Affiliation Computer Science and Systems Engineering Kyushu Institute of Technology()
Date 2015-01-29
Paper # NC2014-59
Volume (vol) vol.114
Number (no) 437
Page pp.pp.-
#Pages 4
Date of Issue