DocumentCode
2970719
Title
Problems of selecting the buttons in the layered neural network application
Author
Kishida, Satoru ; Ishihara, Makoto ; Tokutaka, Heizo ; FujiMura, Kikuo ; Nishimori, Katsumi ; Ishihara, Naganori
Author_Institution
Dept. of Electr. & Electron. Eng., Tottori Univ., Japan
Volume
3
fYear
1993
fDate
25-29 Oct. 1993
Firstpage
2528
Abstract
We applied the back-propagation method of a layered neural network into the problem selecting the buttons, which were fit to the cloth or the thread of suits. The problem is thought to be important as the example of the network which has a number of output units in comparison with the number of input units. The effective method for solving the problem was to increase input units by patterning the input data and to decrease output units by classifying the output. As a result, the square error of the network decreased and the recognition rate of the buttons for the learning patterns improved remarkably.
Keywords
backpropagation; neural nets; pattern recognition; textile industry; back-propagation; button selection; cloth; multilayered neural network application; output classification; square error; suits; thread; Error correction; Intelligent networks; Learning systems; Neural networks; Pattern recognition; Yarn;
fLanguage
English
Publisher
ieee
Conference_Titel
Neural Networks, 1993. IJCNN '93-Nagoya. Proceedings of 1993 International Joint Conference on
Print_ISBN
0-7803-1421-2
Type
conf
DOI
10.1109/IJCNN.1993.714239
Filename
714239
Link To Document