Á¤º¸°úÇÐȸ ³í¹®Áö B : ¼ÒÇÁÆ®¿þ¾î ¹× ÀÀ¿ë
ÇѱÛÁ¦¸ñ(Korean Title) |
ºñ¼±Çü Ư¡ ÃßÃâÀ» À§ÇÑ ¿Â¶óÀÎ ºñ¼±Çü ÁÖ¼ººÐºÐ¼® ±â¹ý |
¿µ¹®Á¦¸ñ(English Title) |
On-line Nonlinear Principal Component Analysis for Nonlinear Feature Extraction |
ÀúÀÚ(Author) |
±èº´ÁÖ
½ÉÁÖ¿ë
ȲâÇÏ
±èÀÏ°ï
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 31 NO. 03 PP. 0361 ~ 0368 (2004. 03) |
Çѱ۳»¿ë (Korean Abstract) |
º» ³í¹®¿¡¼´Â ¿Â¶óÀÎ ÇнÀ ÀÚ·áÀÇ ºñ¼±Çü Ư¡(feature) ÃßÃâÀ» À§ÇÑ »õ·Î¿î ¿Â¶óÀÎ ºñ¼±Çü ÁÖ¼ººÐºÐ¼®(OL-NPCA : On-line Nonlinear Principal Component Analysis) ±â¹ýÀ» Á¦¾ÈÇÑ´Ù. ºñ¼±Çü Ư¡ ÃßÃâÀ» À§ÇÑ ´ëÇ¥ÀûÀÎ ¹æ¹ýÀ¸·Î Ä¿³Î ÁÖ¼ººÐ¹æ¹ý(Kernel PCA)ÀÌ »ç¿ëµÇ°í Àִµ¥ ±âÁ¸ÀÇ Ä¿³Î ÁÖ¼ººÐ ºÐ¼® ¹æ¹ýÀº ´ÙÀ½°ú °°Àº ´ÜÁ¡ÀÌ ÀÖ´Ù. ù° Ä¿³Î ÁÖ¼ººÐ ºÐ¼® ¹æ¹ýÀ» N °³ÀÇ ÇнÀ ÀÚ·á¿¡ Àû¿ëÇÒ ¶§ N¡¿N Å©±âÀÇ Ä¿³Î Çà·ÄÀÇ ÀúÀå ¹× °íÀ¯º¤Å͸¦ °è»ê ÇÏ¿©¾ß Çϴµ¥, NÀÇ Å©±â°¡ Å« °æ¿ì¿¡´Â ¼öÇà¿¡ ¹®Á¦°¡ µÈ´Ù. µÎ ¹ø° ¹®Á¦´Â »õ·Î¿î ÇнÀ ÀÚ·áÀÇ Ãß°¡¿¡ ÀÇÇÑ °íÀ¯°ø°£À» »õ·Î °è»êÇØ¾ß ÇÏ´Â ´ÜÁ¡ÀÌ ÀÖ´Ù. OL-NPCA´Â ÀÌ·¯ÇÑ ¹®Á¦Á¡µéÀ» Á¡ÁøÀûÀÎ °íÀ¯°ø°£ °»½Å ±â¹ý°ú Ư¡ »ç»ó ÇÔ¼ö¿¡ ÀÇÇØ ÇØ°áÇÏ¿´´Ù. Toy µ¥ÀÌŸ¿Í ´ë¿ë·® µ¥ÀÌŸ¿¡ ´ëÇÑ ½ÇÇèÀ» ÅëÇØ OL-NPCA´Â ´ÙÀ½°ú °°Àº ÀåÁ¡À» ³ªÅ¸³½´Ù. ù° ¸Þ¸ð¸® ¿ä±¸·®¿¡ ÀÖ¾î ±âÁ¸ÀÇ Ä¿³Î ÁÖ¼ººÐºÐ¼® ¹æ¹ý¿¡ ºñÇØ »ó´çÈ÷ È¿À²ÀûÀÌ´Ù. µÎ ¹ø° ¼öÇà ¼º´É¿¡ ÀÖ¾î Ä¿³Î ÁÖ¼ººÐ ºÐ¼®°ú À¯»çÇÑ ¼º´ÉÀ» ³ªÅ¸³»¾ú´Ù. ¶ÇÇÑ Á¦¾ÈµÈ OL-NPCA ¹æ¹ýÀº ÀçÇнÀ¿¡ ÀÇÇØ ½±°Ô ¼º´ÉÀÌ Çâ»ó µÇ´Â ÀåÁ¡À» °¡Áö°í ÀÖ´Ù.
|
¿µ¹®³»¿ë (English Abstract) |
The purpose of this study is to propose a new on-line nonlinear PCA(OL-NPCA) method for a nonlinear feature extraction from the incremental data. Kernel PCA(KPCA) is widely used for nonlinear feature extraction, however, it has been pointed out that KPCA has the following problems. First, applying KPCA to N patterns requires storing and finding the eigenvectors of a N¡¿N kernel matrix, which is infeasible for a large number of data N. Second problem is that in order to update the eigenvectors with an another data, the whole eigenspace should be recomputed. OL-NPCA overcomes these problems by incremental eigenspace update method with a feature mapping function. According to the experimental results, which comes from applying OL-NPCA to a toy and a large data problem, OL-NPCA shows following advantages. First, OL-NPCA is more efficient in memory requirement than KPCA. Second advantage is that OL-NPCA is comparable in performance to KPCA. Furthermore, performance of OL-NPCA can be easily improved by re-learning the data.
|
Å°¿öµå(Keyword) |
°íÀ¯º¤ÅÍ
°íÀ¯°ª
Ư¡ »ç»ó ÇÔ¼ö
Ä¿³Î ÁÖ¼ººÐºÐ¼®
¿Â¶óÀÎ ºñ¼±Çü ÁÖ¼ººÐºÐ¼®
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|