Title :
Combined SOM and LVQ based classifiers for handwritten digit recognition
Author :
Wu, Jing ; Yan, Hong
Author_Institution :
Dept. of Electr. Eng., Sydney Univ., NSW, Australia
Abstract :
This paper presents a two-layer self-organizing neural network based technique for handwritten digit recognition. In this method, two classifiers are built with different sets of features using self-organizing map (SOM) and learning vector quantization (LVQ) based algorithms. The two classifiers are then combined to make the final decision. For over 10,000 digit samples which are not used for training extracted from the NIST database, the two classifiers can correctly recognize 97.11% and 97.16% of the digits respectively and the combined classifier has a recognition rate of 98.88%
Keywords :
character recognition; pattern classification; self-organising feature maps; vector quantisation; NIST database; handwritten digit recognition; image classifiers; learning vector quantization; self-organizing map; self-organizing neural network; Handwriting recognition; NIST; Neural networks; Organizing; Robustness; Spatial databases; Testing; Time measurement; Training data; Vector quantization;
Conference_Titel :
Neural Networks, 1995. Proceedings., IEEE International Conference on
Conference_Location :
Perth, WA
Print_ISBN :
0-7803-2768-3
DOI :
10.1109/ICNN.1995.487274
