Presentation 2004/10/12
Analysis of Ensemble Learning for Committee Machine Teacher
Seiji MIYOSHI, Kazuyuki HARA, Masato OKADA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) A major merit of ensemble learning is to realize the input-output relations by combining students that cannot be represented by one student. Therefore, ensemble learning in which a teacher isn't in the model space of one student is very attractive. In this paper ensemble learning, in which a teacher and students are a committee machine and simple perceptrons respectively, is discussed based on online learning theory and statistical mechanics. Hebbian learning gathers all students to the center of teacher units. Perceptron learning keeps a variety of students and the effect of ensemble doesn't disappear. AdaTron learning shows a kind of over-learning.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) ensemble learning / online learning / committee machine / generalization error
Paper # NC2004-79
Date of Issue

Conference Information
Committee NC
Conference Date 2004/10/12(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) Analysis of Ensemble Learning for Committee Machine Teacher
Sub Title (in English)
Keyword(1) ensemble learning
Keyword(2) online learning
Keyword(3) committee machine
Keyword(4) generalization error
1st Author's Name Seiji MIYOSHI
1st Author's Affiliation Kobe City College of Technology()
2nd Author's Name Kazuyuki HARA
2nd Author's Affiliation Tokyo Metropolitan College of Technology
3rd Author's Name Masato OKADA
3rd Author's Affiliation Division of Transdisciplinary Sciences, Graduate School of Frontier Sciences, The University of Tokyo:RIKEN Brain Science Institute:JST PRESTO
Date 2004/10/12
Paper # NC2004-79
Volume (vol) vol.104
Number (no) 349
Page pp.pp.-
#Pages 6
Date of Issue