• Àüü
  • ÀüÀÚ/Àü±â
  • Åë½Å
  • ÄÄÇ»ÅÍ
´Ý±â

»çÀÌÆ®¸Ê

Loading..

Please wait....

±¹³» ÇÐȸÁö

Ȩ Ȩ > ¿¬±¸¹®Çå > ±¹³» ÇÐȸÁö > µ¥ÀÌÅͺ£À̽º ¿¬±¸È¸Áö(SIGDB)

µ¥ÀÌÅͺ£À̽º ¿¬±¸È¸Áö(SIGDB)

Current Result Document :

ÇѱÛÁ¦¸ñ(Korean Title) Multiple Exponential HistogramÀ» »ç¿ëÇÑ °³³ä º¯È­ °ËÃâ ±â¹ý
¿µ¹®Á¦¸ñ(English Title) A Method for Detecting Concept Drift by Using Multiple Exponential Histogram
ÀúÀÚ(Author) ±è¸¸¼ö   ÀÓÈ¿»ó   ManSoo Kim   Hyo-Sang Lim  
¿ø¹®¼ö·Ïó(Citation) VOL 34 NO. 01 PP. 0039 ~ 0058 (2018. 04)
Çѱ۳»¿ë
(Korean Abstract)
º» ³í¹®Àº µ¥ÀÌÅͽºÆ®¸² ȯ°æ¿¡¼­ °³³ä º¯È­¸¦ °ËÃâÇÏ´Â ¹æ¹ýÀ» Á¦¾ÈÇÑ´Ù. ¸ÕÀú µ¥ÀÌÅͽºÆ®¸²¿¡¼­ÀÇ °³³äÀ» Åë°èÀû Ư¡ÀÎ ÀÔ·Â µ¥ÀÌÅÍÀÇ ºÐÆ÷·Î ¸ðµ¨¸µÇÏ°í, ´ÙÀ½°ú °°Àº ¹æ¹ýÀ» »ç¿ëÇÏ¿© °³³ä º¯È­¸¦ ŽÁöÇÑ´Ù: 1) µ¥ÀÌÅͽºÆ®¸²¿¡ ½½¶óÀ̵ù À©µµ¿ì¸¦ »ç¿ëÇÏ¿© ½Ç½Ã°£ »ðÀÔ µ¥ÀÌÅ͸¦ ¸ðµ¨¸µ ÇÑ´Ù. 2) °¢ À©µµ¿ì¿¡ ´ëÇÑ È÷½ºÅä±×·¥À» ±¸ÃàÇÏ°í, 3) ±¸Ãà µÈ È÷½ºÅä±×·¥ÀÇ º¯È­¸¦ ÃøÁ¤ÇÏ¿© °³³äº¯È­ ¿©ºÎ¸¦ ÆÇ´ÜÇÑ´Ù. À̶§, µ¥ÀÌÅͽºÆ®¸²ÀÇ ½Ç½Ã°£¼ºÀ» °í·ÁÇÏ¿© È÷½ºÅä±×·¥À» ±¸ÃàÇϱâ À§Çؼ­ 0°ú 1·Î ÀÌ·ç¾îÁø µ¥ÀÌÅ͸¦ ŸÀÓ½ºÅÆÇÁ¿Í ÇÔ²² ±Ù»çÈ­ ÇÏ¿© ÀúÀåÇÏ´Â µ¥ÀÌÅÍ ±¸Á¶ÀÎ Exponential Histogram(EH)À» »ç¿ëÇÑ´Ù. ¶ÇÇÑ ±Ù»çÈ­ ±¸Á¶ÀÎ EHÀÇ ¿ÀÂ÷¸¦ °í·ÁÇÏ¿© º¸´Ù Á¤È®ÇÏ°Ô ºÐÆ÷¸¦ ºñ±³ÇÒ ¼ö ÀÖ´Â ¹æ¹ýÀ» Á¦¾ÈÇÑ´Ù.
¿µ¹®³»¿ë
(English Abstract)
In this paper, we propose a method for detecting concept drifts in data streams. We first model the concept as statistical characteristics, specifically as distributions of input data, and then, use the following method to detect concept drifts (i.e., changes of the data distribution) in data stream: 1) we model real-time data streams as sliding windows, 2) build a histogram for each window, and 3) detect concept drifts by detecting changes in the histograms. In order to solve the real-time feature of data streams, we exploit Exponential Histogram (EH) which is a data structure that approximately stores data consisting of 0s and 1s with their timestamps. We also provide a distribution comparison method that considers the error of EH in order to improve the accuracy of the concept drift detection.
Å°¿öµå(Keyword) µ¥ÀÌÅͽºÆ®¸²   °³³ä º¯È­   Exponential Histogram   Data Stream   Concept Drift   Exponential Histogram  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå