Generalized Learning Vector Quantization

We propose a new learning method, "Generalized Learning Vector Quantization (GLVQ)," in which reference vectors are updated based on the steepest descent method in order to minimize the cost function. The cost function is determined so that the obtained learning rule satisfies the convergence condition. We prove that Kohonen's rule as used in LVQ does not satisfy the convergence condition and thus degrades recognition ability. Experimental results for printed Chinese character recognition reveal that GLVQ is superior to LVQ in recognition ability.

[1]  K. Yamada,et al.  On-line Japanese character recognition experiments by an off-line method based on normalization-cooperated feature extraction , 1993, Proceedings of 2nd International Conference on Document Analysis and Recognition (ICDAR '93).

[2]  Teuvo Kohonen,et al.  Self-Organization and Associative Memory , 1988 .

[3]  Seong-Whan Lee,et al.  Optimal design of reference models using simulated annealing combined with an improved LVQ3 , 1993, Proceedings of 2nd International Conference on Document Analysis and Recognition (ICDAR '93).

[4]  Lawrence D. Jackel,et al.  Handwritten Digit Recognition with a Back-Propagation Network , 1989, NIPS.

[5]  Teuvo Kohonen,et al.  Self-organization and associative memory: 3rd edition , 1989 .

[6]  A. Sato,et al.  A criterion for training reference vectors and improved vector quantization , 1994, Proceedings of 1994 IEEE International Conference on Neural Networks (ICNN'94).

[7]  James C. Bezdek,et al.  Generalized clustering networks and Kohonen's self-organizing scheme , 1993, IEEE Trans. Neural Networks.

[8]  Manuel Graña,et al.  An analysis of the GLVQ algorithm , 1995, IEEE Trans. Neural Networks.

[9]  K. Yamada,et al.  Handwritten numeral recognition by multilayered neural network with improved learning algorithm , 1989, International 1989 Joint Conference on Neural Networks.

[10]  Fumio Yoda,et al.  Printed Japanese character recognition based on multiple modified LVQ neural network , 1993, Proceedings of 2nd International Conference on Document Analysis and Recognition (ICDAR '93).