• 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 : 9 / 18 ÀÌÀü°Ç ÀÌÀü°Ç   ´ÙÀ½°Ç ´ÙÀ½°Ç

ÇѱÛÁ¦¸ñ(Korean Title) À̵¿°æ·Î ¿¹ÃøÀ» À§ÇÑ µµ·Î ³×Æ®¿öÅ© ÅäÆú·ÎÁö ±â¹Ý µö ±×¸®µå ÀÓº£µù
¿µ¹®Á¦¸ñ(English Title) Deep Grid Embedding using a Road Network Topology for Route Prediction
ÀúÀÚ(Author) °­ÁØÇõ   Junhyeok Kang   ÀÌÀç±æ   Jae-Gil Lee   ±èº´Áø   Byeong-jin Kim   À̼ö¿ø   Suwon Lee  
¿ø¹®¼ö·Ïó(Citation) VOL 36 NO. 01 PP. 0057 ~ 0073 (2020. 04)
Çѱ۳»¿ë
(Korean Abstract)
À̵¿°æ·Î ¿¹ÃøÀº ¿òÁ÷ÀÌ´Â °´Ã¼°¡ Áö³ª¿Â °æ·Î·ÎºÎÅÍ ´ÙÀ½ À§Ä¡¸¦ ¿¹ÃøÇÏ´Â ÀÛ¾÷À¸·Î À̵¿°æ·Î µ¥ÀÌÅ͸¶ÀÌ´×ÀÇ ÁÖ¿ä ºÐ¾ß Áß ÇϳªÀÌ´Ù. À̵¿°æ·Î¸¦ ¿¹ÃøÇϱâ À§Çؼ­´Â ÇöÀç À§Ä¡·ÎºÎÅÍ µµ´Þ°¡´ÉÇÑ ´ÙÀ½ À§Ä¡µéÀ» ÆľÇÇÏ´Â °ÍÀÌ Áß¿äÇÏ´Ù. µµ·Î ³×Æ®¿öÅ©¿¡´Â ´Ù¾çÇÑ ÇüÅÂÀÇ ÅäÆú·ÎÁö°¡ Á¸ÀçÇϸç À̵¿°æ·Î´Â µµ·Î ³×Æ®¿öÅ©ÀÇ ÅäÆú·ÎÁö¿¡ µû¶ó Á¦¾àÀûÀ¸·Î »ý¼ºµÈ´Ù. ÀÌ·¯ÇÑ µµ·Î ³×Æ®¿öÅ©ÀÇ ÅäÆú·ÎÁö Á¤º¸´Â ÇöÀç À§Ä¡·ÎºÎÅÍ µµ´Þ °¡´ÉÇÑ ´ÙÀ½ À§Ä¡µé¿¡ ´ëÇÑ ³»¿ë¸¦ Æ÷ÇÔÇÏ°í ÀÖ¾î À̸¦ À̵¿°æ·Î ¿¹Ãø ¸ðµ¨¿¡ È°¿ëÇÏ¸é ¿¹Ãø ¼º´ÉÀ» Çâ»ó½Ãų¼ö ÀÖ´Ù. À̵¿°æ·Î µ¥ÀÌÅ͸¶ÀÌ´×À» À§ÇØ ±×¸®µå ÀÓº£µùÀ» ¼öÇàÇÑ ±âÁ¸ ¿¬±¸´Â ÀÓº£µù °úÁ¤¿¡¼­ ±×¸®µåµé °£ÀÇ °ø°£Àû ±ÙÁ¢¼ºÀ» °í·ÁÇÏ¿´´Ù. ÇÏÁö¸¸, ÀÌ´Â °ø°£Àû °Å¸®´Â °¡±õÁö¸¸ µµ´ÞÇÒ ¼ö ¾ø´Â Áö¿ªÀ» ÀáÀç °ø°£ ³»¿¡ ¼­·Î °¡±õ°Ô ÀÓº£µùÇϰԵǴ ¹®Á¦°¡ ¹ß»ýÇϱ⠶§¹®¿¡ À̵¿°æ·Î ¿¹Ãø ¹®Á¦¿¡ ÀûÇÕÇÏÁö ¾Ê´Ù. º» ¿¬±¸¿¡¼­´Â µµ·Î ³×Æ®¿öÅ©ÀÇ ÅäÆú·ÎÁö¸¦ °í·ÁÇÏ´Â ±×¸®µå ÀÓº£µù ¹æ¹ýÀ» Á¦¾ÈÇÑ´Ù. Á¦¾ÈÇÏ´Â ¹æ¹ý·ÐÀº °ø°£Àû ±ÙÁ¢¼º°ú ¿¬°á¼ºÀ» ±â¹ÝÀ¸·Î ±×¸®µåµé °£ÀÇ ÅäÆú·ÎÁöÀû À¯»çµµ¸¦ Á¤ÀÇÇÏ°í ¼­·Î À¯»çÇÑ ±×¸®µåµéÀ» ÀáÀç °ø°£¿¡¼­ ¼­·Î °¡±õ°Ô ÀÓº£µùÇÏ¿© À̵¿°æ·Î ¿¹Ãø ¼º´ÉÀ» Çâ»ó½ÃŲ´Ù. ½ÇÁ¦ Åýà À̵¿°æ·Î µ¥ÀÌÅ͸¦ È°¿ëÇÑ À̵¿°æ·Î ¿¹Ãø ½ÇÇè¿¡¼­ Á¦¾ÈÇÑ ¹æ¹ý·ÐÀÌ ±âÁ¸ÀÇ ¹æ¹ýµéº¸´Ù ³ôÀº Á¤È®µµ¸¦ º¸¿´´Ù.
¿µ¹®³»¿ë
(English Abstract)
Route prediction is one of the main areas of trajectory data mining, which is to predict the next location of a moving object from the partial historical trajectory. To predict the next location of the moving object, it is important to know the next locations reachable from the current location. A road network has various types of topologies, and the trajectories are generated constrained by the topology of the road network. Using topology information helps a prediction model to get better route prediction performance because it lets the model know the next locations reachable from the current location. Existing methods of grid embedding for trajectory data mining consider mostly spatial proximity between two grids during the embedding process. However, this approach is not suitable for the route prediction because it embeds the locations which are spatially close but unreachable as similar representations into latent space. In this paper, we propose a grid embedding method considering the topology of road networks, which defines topological similarities between grids based on their spatial proximity and connectivity. As a result, our method improves route prediction performance by embedding topologically similar grids close together in a latent space. Experiments on real taxi trajectory data demonstrate that our method improves accuracy compared with existing methods.
Å°¿öµå(Keyword) À̵¿°æ·Î ¿¹Ãø   ±×¸®µå ÀÓº£µù   ³×Æ®¿öÅ© ÅäÆú·ÎÁö   Route Prediction   Grid Embedding   Network Topology  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå