DocumentCode
324550
Title
Neocognitron with improved bend-extractors
Author
Fukushima, Kunihiko ; Kimura, Eiji ; Shouno, Hayaru
Author_Institution
Graduate Sch. of Eng. Sci., Osaka Univ., Japan
Volume
2
fYear
1998
fDate
4-9 May 1998
Firstpage
1172
Abstract
We (1988) have reported previously that the performance of a neocognitron can be improved by a built-in bend-extracting layer. The conventional bend-extracting layer can detect bend points and end points of lines correctly, but not always crossing points of lines. This paper discusses that an introduction of a mechanism of disinhibition can make the bend-extracting layer detect not only bend points and end points but also crossing points of lines correctly. A neocognitron with this improved bend-extracting layer can recognize handwritten digits in the real world with a recognition rate of 98%
Keywords
character recognition; edge detection; feature extraction; feedforward neural nets; bend points; bend-extracting layer; crossing points; end points; feature extraction; handwritten digit recognition; line detection; multilayer neural networks; neocognitron; Data mining; Handwriting recognition; Pattern recognition; Robustness;
fLanguage
English
Publisher
ieee
Conference_Titel
Neural Networks Proceedings, 1998. IEEE World Congress on Computational Intelligence. The 1998 IEEE International Joint Conference on
Conference_Location
Anchorage, AK
ISSN
1098-7576
Print_ISBN
0-7803-4859-1
Type
conf
DOI
10.1109/IJCNN.1998.685939
Filename
685939
Link To Document