Á¤º¸°úÇÐȸ ÄÄÇ»ÆÃÀÇ ½ÇÁ¦ ³í¹®Áö (KIISE Transactions on Computing Practices)
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
¾ÏÈ£È ¿¬»ê ÇÁ·ÎÅäÄÝ ¹× µ¥ÀÌÅÍ ºÐÆ÷ ±â¹Ý Áß½ÉÁ¡ ¼±Á¤À» ÀÌ¿ëÇÑ °³ÀÎ Á¤º¸ º¸È£ k-Means Ŭ·¯½ºÅ͸µ ¾Ë°í¸®Áò |
¿µ¹®Á¦¸ñ(English Title) |
Privacy-preserving k-Means Clustering Algorithm using a Secure Comparison Operation Protocol and Data Distribution-based Center Selection |
ÀúÀÚ(Author) |
±èÇüÁø
ÀåÀç¿ì
Hyeong-Jin Kim
Jae-Woo Chang
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 24 NO. 10 PP. 0505 ~ 0512 (2018. 10) |
Çѱ۳»¿ë (Korean Abstract) |
¾Æ¿ô¼Ò½Ì µ¥ÀÌÅͺ£À̽º »ó¿¡¼ °³ÀÎÁ¤º¸¸¦ º¸È£ Ŭ·¯½ºÅ͸µ ¾Ë°í¸®ÁòÀÌ È°¹ßÈ÷ ¿¬±¸µÇ¾ú´Ù. Rao. et. al. Àº paillier ¾ÏÈ£È ½Ã½ºÅÛÀ» ÀÌ¿ëÇÏ¿© Á¤º¸ º¸È£¸¦ Áö¿øÇÏ´Â k-Means Ŭ·¯½ºÅ͸µ ¾Ë°í¸®ÁòÀ» Á¦¾ÈÇÏ¿´´Ù. ±×·¯³ª ÇØ´ç ¾Ë°í¸®ÁòÀº Ãʱâ Áß½ÉÁ¡À» ÀÓÀÇ·Î ¼±Á¤ÇÔÀ¸·Î½á ClusteringÀÇ °á°ú°¡ ºÒ±ÔÄ¢ÇÏ´Ù. ¾Æ¿ï·¯, ºñÆ® ¹è¿ ±â¹Ý ºñ±³ ¿¬»êÀÚ¸¦ »ç¿ëÇϱ⠶§¹®¿¡ ¹è¿ÀÇ Å©±â¿¡ ºñ·ÊÇÏ¿© °è»ê ºñ¿ëÀÌ Å©°Ô Áõ°¡ÇÏ´Â ´ÜÁ¡ÀÌ Á¸ÀçÇÑ´Ù. ÀÌ·¯ÇÑ ¹®Á¦Á¡À» ÇØ°áÇϱâ À§ÇØ. º» ³í¹®¿¡¼´Â Á¤º¸ º¸È£¸¦ Áö¿øÇÏ´Â È¿À²ÀûÀÎ k-Means Ŭ·¯½ºÅ͸µ ¾Ë°í¸®ÁòÀ» Á¦¾ÈÇÑ´Ù. À̸¦ À§ÇØ, ¾ÏÈ£ÈµÈ µ¥ÀÌÅÍÀÇ ºñ±³¸¦ ¼öÇàÇÏ´Â ¾ÏÈ£È ¿¬»ê ÇÁ·ÎÅäÄÝÀ» Á¦¾ÈÇÑ´Ù. ¶ÇÇÑ Àüü µ¥ÀÌÅÍ ºÐÆ÷¸¦ ¹Ý¿µÇÏ¿© Ãʱâ Áß½ÉÁ¡À» ¼±Á¤ÇÔÀ¸·Î½á È¿À²ÀûÀΠŬ·¯½ºÅ͸µÀ» ¼öÇàÇÑ´Ù. ¸¶Áö¸·À¸·Î, ¼º´ÉÆò°¡¸¦ ÅëÇØ Á¦¾ÈÇÏ´Â ¾Ë°í¸®ÁòÀÌ ±âÁ¸ ¾Ë°í¸®Áò¿¡ ºñÇØ Æò±ÕÀûÀ¸·Î ¾à 150¢¦250% ¼º´ÉÀÌ Çâ»óµÊÀ» º¸ÀδÙ.
|
¿µ¹®³»¿ë (English Abstract) |
Privacy-preserving clustering algorithms in outsourced databases have been actively studied. Rao. et. al. proposed a k-Means clustering algorithm that supports the protection of information by using a paillier crypto system. However, since the algorithm selects initial center points randomly, the result of clustering is irregular. Because it uses a comparison operator based on bit arrays, it has a disadvantage in that its computation cost greatly increases in proportion to the size of the array. To solve this problem, we propose an efficient k-Means clustering algorithm that supports the protection of information. To do this, we propose a cryptographic operation protocol that compares encrypted data. In addition, we performs efficient clustering by selecting initial center points in terms of the entire data distribution. Finally, we show through our performance evaluation that the proposed algorithm outperforms the existing algorithm by 150 to 250% on the average.
|
Å°¿öµå(Keyword) |
¾Æ¿ô¼Ò½Ì µ¥ÀÌÅͺ£À̽º
¾ÏÈ£È µ¥ÀÌÅͺ£À̽º
k-Means Ŭ·¯½ºÅ͸µ ¾Ë°í¸®Áò
¾ÏÈ£È µ¥ÀÌÅÍ ¸¶ÀÌ´×
Á¢±Ù ÆÐÅÏ º¸È£
database outsourcing
database encryption
k-Means clustering
encrypted data mining
hiding data access patterns
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|