Presentation 1995/7/20
High Accuracy Recognition of ETL9B using Exclusive Learning Neural Network-II (ELNET-II)
Kazuki Saruta, Ning Sun, Masato Abe, Yoshiaki Nemoto,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We have already proposed Exclusive Learning neural NETwork (ELNET) as neural network which can be easily utilized to construct a recognition system for Chinese character. In this paper, we propose ELNET-II, which is improved version of ELNET and use more effective learning algorithm. In ELNET, some problem such effect of learning remained. In ELNET-II, the number of learning data and module size variable according to candidate appearance frequency in rough classification. In recognition experiment for ETL9B (3036 categories), we have gotten 95.84% as maximum recognition rate. In comparison with ELNET, the rate improved about 1.3%.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) handwritten character recognition / neural networks / ETL9B / ELNET-II
Paper #
Date of Issue

Conference Information
Committee PRU
Conference Date 1995/7/20(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 Pattern Recognition and Understanding (PRU)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) High Accuracy Recognition of ETL9B using Exclusive Learning Neural Network-II (ELNET-II)
Sub Title (in English)
Keyword(1) handwritten character recognition
Keyword(2) neural networks
Keyword(3) ETL9B
Keyword(4) ELNET-II
1st Author's Name Kazuki Saruta
1st Author's Affiliation Graduate School of Information Sciences, Tohoku Univ.()
2nd Author's Name Ning Sun
2nd Author's Affiliation Graduate School of Information Sciences, Tohoku Univ.
3rd Author's Name Masato Abe
3rd Author's Affiliation Computer Center, Tohoku Univ.
4th Author's Name Yoshiaki Nemoto
4th Author's Affiliation Graduate School of Information Sciences, Tohoku Univ.
Date 1995/7/20
Paper #
Volume (vol) vol.95
Number (no) 164
Page pp.pp.-
#Pages 6
Date of Issue