Á¤º¸°úÇÐȸ ³í¹®Áö B : ¼ÒÇÁÆ®¿þ¾î ¹× ÀÀ¿ë
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
NSGA-II ¾Ë°í¸®ÁòÀ» ÀÌ¿ëÇÑ ´ÙÁß ·¹ÀÌºí ºÐ·ù ¹®Á¦¿¡¼ Ư¡ ¼±º° |
¿µ¹®Á¦¸ñ(English Title) |
Feature Selection in Multi-label Classification using NSGA-II Algorithm |
ÀúÀÚ(Author) |
À±Á¤ÈÆ
ÀÌÀ缺
±è´ë¿ø
Jeonghun Yoon
Jaesung Lee
Dae-Won Kim
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 40 NO. 03 PP. 0133 ~ 0140 (2013. 03) |
Çѱ۳»¿ë (Korean Abstract) |
ÃÖ±Ù Çϳª ÀÌ»óÀÇ Å¬·¡½º ·¹À̺íÀ» °¡Áö´Â µ¥ÀÌÅÍ¿¡ ´ëÇÑ Å¬·¡½ººÐ·ù ±â¹ýµéÀÌ ¿¬±¸µÇ°í ÀÖ´Ù. ±× Áß ¸î¸î ¿¬±¸µé¿¡¼ ´ÙÁß ·¹ÀÌºí ºÐ·ùÀÇ ¼º´ÉÀ» ³ôÀ̱â À§ÇØ Æ¯Â¡ ¼±º° ±â¹ýÀ» »ç¿ëÇß´Ù. ±×·¯³ª ´ÙÁß ·¹ÀÌºí µ¥ÀÌÅÍÀÇ º¹À⼺À¸·Î ÀÎÇØ ±âÁ¸ÀÇ Æ¯Â¡ ¼±º° ±â¹ýÀ» Àû¿ëÇÏ´Â °ÍÀº ¼º´É Çâ»ó¿¡ ÇÑ°è°¡ ÀÖ´Ù. º» ³í¹®Àº ´Ù¸ñÀû ÃÖÀûÈ ¾Ë°í¸®ÁòÀÎ NSGA-II¸¦ ´ÙÁß ·¹À̺í Ư¡ ¼±º° ¹®Á¦¿¡ È°¿ëÇß´Ù. ¶ÇÇÑ º» ³í¹®¿¡¼´Â Ư¡ ¼±º° ¹®Á¦¿¡ ÀûÇÕÇÑ À¯ÀüÀÚ Á¶ÀÛ ±â¹ýÀ» Á¦¾ÈÇÏ¿© NSGA-II¿¡ Àû¿ëÇß´Ù. ±×¸®°í ½ÇÁ¦ ´ÙÁß ·¹ÀÌºí µ¥ÀÌÅÍ ¼Â¿¡¼ ½ÇÇèÀ» ÅëÇØ Á¦¾ÈÇÏ´Â ¾Ë°í¸®ÁòÀÌ ±âÁ¸ À¯ÀüÀÚ ¾Ë°í¸®ÁòÀ» »ç¿ëÇÑ Æ¯Â¡ ¼±º° ±â¹ýº¸´Ù ´õ ÁÁÀº ¼º´ÉÀ» °¡Áö´Â °ÍÀ» º¸ÀδÙ.
|
¿µ¹®³»¿ë (English Abstract) |
Recently, a lot of researchers are interested in multi-label classification. Some of the researchers use feature subset selection to improve performance in multi-label classification. However, because multi-label problem is more complex than single-label, single-label feature selection algorithms have limitation to be applied to multi-label data. In this paper, NSGA-II, which is multi-objective optimize algorithm, is used for multi-label feature selection. In addition, this study proposes a novel genetic operator for feature selection problem. Finally, experiments on real world data set show that proposed algorithm achieves better performance than traditional genetic algorithm.
|
Å°¿öµå(Keyword) |
´ÙÁß ·¹ÀÌºí ºÐ·ù
Ư¡ ¼±º° ±â¹ý
´Ù¸ñÀû À¯ÀüÀÚ ¾Ë°í¸®Áò
multi-label classification
feature selection
NSGA-II
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|