Presentation 2002/5/13
A Note on Computational Learning and Information Compression
Makoto NAKAZAWA, Toshiyasu MATSUSHIMA, Shigeichi HIRASAWA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Recently, grammar based codes are researched in the area of lossless source coding. But, it is not clear which formal grammars is appropriate in Chomsky hierarchy for information compression. In this paper, we consider grammar based codes as the process of deriving automata based on machine learning and it is a purpose to clarify what influence the class of grammars or languages gives the encoding in the view of machine learning. I will show that Bayes algorithm outputs the shortest hypothesis in the class and some class has feasible complexity in order to ouput optimal hypothesis based on computational learning theory.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) formal grammars / lossless compression / computational learning / Bayes algorithm / time complexity
Paper # IT2002-7
Date of Issue

Conference Information
Committee IT
Conference Date 2002/5/13(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 Information Theory (IT)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A Note on Computational Learning and Information Compression
Sub Title (in English)
Keyword(1) formal grammars
Keyword(2) lossless compression
Keyword(3) computational learning
Keyword(4) Bayes algorithm
Keyword(5) time complexity
1st Author's Name Makoto NAKAZAWA
1st Author's Affiliation Media Network Center, Waseda University()
2nd Author's Name Toshiyasu MATSUSHIMA
2nd Author's Affiliation School of Science and Engineering, Waseda University
3rd Author's Name Shigeichi HIRASAWA
3rd Author's Affiliation School of Science and Engineering, Waseda University
Date 2002/5/13
Paper # IT2002-7
Volume (vol) vol.102
Number (no) 66
Page pp.pp.-
#Pages 6
Date of Issue