Title :
A criterion for training reference vectors and improved vector quantization
Author :
SATo, Atsushi ; Tsukumo, Jun
Author_Institution :
C&C Inf. Technol. Res. Labs., NEC Corp., Kawasaki, Japan
fDate :
27 Jun-2 Jul 1994
Abstract :
In this paper, the criterion for training reference vectors is formulated in which the reference vectors are modified by the input vectors closer to decision boundaries. The authors present an improved vector quantization method, based on the above idea. Decision boundaries determined by this method are discussed and it is shown that the proposed method has several advantages as compared with conventional LVQ2. Experimental results for printed Japanese Hiragana characters recognition reveal that the proposed method is superior to LVQ2 and MLP in recognition ability
Keywords :
character recognition; learning (artificial intelligence); neural nets; vector quantisation; decision boundaries; learning algorithm; neural networks; printed Japanese Hiragana characters recognition; training reference vectors; vector quantization; Artificial neural networks; Character recognition; Euclidean distance; Information technology; Large-scale systems; National electric code; Nearest neighbor searches; Pattern recognition; Speech recognition; Vector quantization;
Conference_Titel :
Neural Networks, 1994. IEEE World Congress on Computational Intelligence., 1994 IEEE International Conference on
Conference_Location :
Orlando, FL
Print_ISBN :
0-7803-1901-X
DOI :
10.1109/ICNN.1994.374156
