Á¤º¸°úÇÐȸ ³í¹®Áö B : ¼ÒÇÁÆ®¿þ¾î ¹× ÀÀ¿ë
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
RBF ½Å°æ¸Á ºÐ·ù±âÀÇ È¿À²Àû ±¸¼º ¹æ¹ý |
¿µ¹®Á¦¸ñ(English Title) |
An Efficient Method to Construct a Radial Basis Function Neural Network Classifier |
ÀúÀÚ(Author) |
Ȳ¿µ¼·
¹æ½Â¾ç
Young-Sup Hwang
Sung-Yang Bang
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 24 NO. 05 PP. 0451 ~ 0460 (1997. 05) |
Çѱ۳»¿ë (Korean Abstract) |
RBF ½Å°æ¸ÁÀº ÀÓÀÇÀÇ ÇÔ¼ö¸¦ ±Ù»çÇÒ ¼ö ÀÖÁö¸¸ ÁÖ¾îÁø ¹®Á¦¸¦ Ç®±â À§ÇØ RBF ½Å°æ¸ÁÀ» ±¸¼ºÇÏ´Â °£´ÜÇÑ ¹æ¹ýÀÌ ¾ø¾ú´Ù. ÀÌ ³í¹®Àº RBF ½Å°æ¸ÁÀ» ºÐ·ù±â·Î »ç¿ëÇÒ ¶§ À̸¦ È¿À²ÀûÀÌ°í È¿°úÀûÀ¸·Î ±¸¼ºÇÏ´Â ¹æ¹ýÀ» ¼³¸íÇÑ´Ù. Á¦¾ÈÇÏ´Â ¹æ¹ýÀº ºü¸¥ Ŭ·¯½ºÅ͸µ ¹æ¹ýÀÎ APC-¥²À» ½á¼ Áß°£ÃþÀ» °áÁ¤ÇÏ°í Áß°£Ãþ°ú Ãâ·ÂÃþ »çÀÌÀÇ °¡ÁßÄ¡¸¦ Åë°èÀû ¹æ¹ýÀ¸·Î ±¸ÇÑ´Ù. Á¦¾ÈÇÑ ¹æ¹ýÀ» ¹«Á¦¾à Çʱ⠼ýÀÚ ÀνÄÀ» À§ÇÑ ºÐ·ù±â¸¦ ±¸¼ºÇϴµ¥ Àû¿ëÇÏ¿´´Ù. ½ÇÇèÀº Á¦¾ÈÇÑ ¹æ¹ýÀÌ RBF ½Å°æ¸Á ºÐ·ù±â¸¦ ºü¸£°Ô ±¸¼ºÇÒ ¼ö ÀÖ°í, °°Àº µ¥ÀÌÅͺ£À̽º¸¦ »ç¿ëÇÑ ÀÌÀü ¿¬±¸ °á°úº¸´Ù ºÐ·ù±âÀÇ ¼º´ÉÀÌ ´õ ³ªÀ½À» º¸¿© ÁÖ¾ú´Ù.
|
¿µ¹®³»¿ë (English Abstract) |
Radial basis function(RBF) neural networks have the power of the universal function approximation. But it is usually not straightforward how to construct an RBF neural network to solve a given problem. This paper describes a method to construct an RBF neural network classifier efficiently and effectively. The method determines the middle layer neurons by a fast clustering algorithm, APC-¥² and computes the optimal weights between the middle and the output layers statistically. We applied the proposed method to construct an RBF neural network classifier for an unconstrained handwritten digit recognition. The experiment showed that the method could construct an RBF neural network classifier fast and the performance of the classifier was better than the best result previously reported.
|
Å°¿öµå(Keyword) |
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|