Title :
Competitive self-organizing neural network
Author :
Imoto, Masmichi Mo
Author_Institution :
Basic Res. Dept., Olympus Opt. Co. Ltd., Tokyo, Japan
Abstract :
Employing a network with the energy function which shows a degree of self-organization, learning and association of alphabet letters is shown. Due to the following two improvements introduced in comparison with the precedents, the efficiency of calculation has become much higher: (i) learning of local features of patterns; (ii) putting labels to patterns so as to show the total features of a picture.
Keywords :
pattern recognition; self-organising feature maps; alphabet letters; association; competitive self-organizing neural network; efficiency of calculation; energy function; labels; learning; local features; patterns; Cities and towns; Convergence; Joining processes; Neural networks; Optical computing; Pattern analysis; Performance analysis; Shape;
Conference_Titel :
Neural Networks, 1993. IJCNN '93-Nagoya. Proceedings of 1993 International Joint Conference on
Print_ISBN :
0-7803-1421-2
DOI :
10.1109/IJCNN.1993.716744
