• 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) ÀüÀå »óȲÀ» ¹¦»çÇÏ´Â °¡¼³ °£ °ü·Ã¼ºÀ» ½Äº°Çϱâ À§ÇÑ °èÃþÀû ±×·¡ÇÁ ½Å°æ¸Á
¿µ¹®Á¦¸ñ(English Title) Hierarchical Graph Neural Network for Identifying Relevance between Hypotheses that Describe Battlefield Situation
ÀúÀÚ(Author) ³ªÁöÇý   °­ÁØÇõ   ÀÌÀç±æ   ÀÌ¿ì½Å   Áø¼Ò¿¬   ±è»ó¹Î   Jihye Na   Junhyeok Kang   Jae-Gil Lee   Woosin Lee   Soyeon Jin   Sangmin Kim   Á¶Çö¼ö   ÁÖÇöÁø   Áø¼Ò¿¬   ½ÅÀ¯°æ   ÀÌ¿ì½Å   ½Å±âÁ¤   Hyeonsoo Jo   Hyunjin Choo   Soyeon Jin   Yukyung Shin   Woosin Lee   Kijung Shin  
¿ø¹®¼ö·Ïó(Citation) VOL 38 NO. 02 PP. 0018 ~ 0028 (2023. 01)
Çѱ۳»¿ë
(Korean Abstract)
ÁöÈÖ°üÀÇ ÀÇ»ç°áÁ¤À» Áö¿øÇÏ´Â Áö´ÉÇü ÁöÈÖÅëÁ¦ ü°èÀÇ Á߿伺ÀÌ ³ô¾ÆÁö°í ÀÖ´Ù. À̸¦ À§Çؼ­´Â, ÀüÀå »óȲ Á¤º¸¿¡ ´ëÇÑ Áö½Äº£À̽º ±¸ÃàÀÌ ÇÊ¿äÇѵ¥, ÃÖ±Ù ¿¬±¸¿¡¼­ ÀüÀå »óȲ ºÐ¼®¿ë °¡¼³ µ¥ÀÌÅÍ »ý¼º ¹æ¹ýÀÌ Á¦¾ÈµÇ¾ú´Ù. ÁöÈÖ°üÀÇ ÀÇ»ç°áÁ¤ Áö¿øÀ» À§Çؼ­´Â ÁöÈÖ°üÀÇ ¿ä±¸³ª Áú¹®¿¡ °ü·ÃµÈ Áö½Ä ¿ä¼Ò¸¦ ÃßÃâÇÒ ¼ö ÀÖ¾î¾ß ÇÏ°í, À̸¦ À§Çؼ­´Â °¡¼³ °£ÀÇ °ü·Ã¼ºÀ» ÆľÇÇÏ¿© °ü·ÃµÈ Á¤º¸¸¦ ÃßÃâÇÏ¿©¾ß ÇÑ´Ù. ÇÏÁö¸¸, °¡¼³ ³»ÀÇ Áö½Ä ¿ä¼ÒµéÀº ¼­·Î °èÃþÀûÀ¸·Î ¿¬°áµÇ¾î ÀÖ¾î ÀϹÝÀûÀÎ ½Å°æ¸ÁÀ¸·Î´Â °¡¼³ °£ÀÇ °ü·Ã¼ºÀ» Æò°¡ÇÏ´Â °ÍÀÌ ¾î·Æ´Ù. º» ³í¹®¿¡¼­´Â °¡¼³ÀÇ °èÃþÀûÀÎ ±¸Á¶¸¦ È°¿ëÇÏ¿© °¡¼³ ³»ÀÇ ÀüÀå Áö½Ä ¿ä¼ÒµéÀ» È¿°úÀûÀ¸·Î Áý°èÇÏ°í, À̸¦ ±â¹ÝÀ¸·Î µÎ °¡¼³ÀÇ °ü·Ã¼ºÀ» ÆÇ´ÜÇÒ ¼ö ÀÖ´Â °èÃþÀû ±×·¡ÇÁ ½Å°æ¸Á »óȲ ½Äº° ÇнÀ¸ðµ¨À» ¼³°èÇÏ¿´´Ù. ¶ÇÇÑ, Á¦¾ÈµÈ ÇнÀ¸ðµ¨ÀÌ ±âº» ÇнÀ¸ðµ¨º¸´Ù Á¤È®µµ°¡ ¿ì¼öÇϸç, ƯÈ÷ µ¥ÀÌÅÍÀÇ 40%¸¸ ÇнÀ¿¡ »ç¿ëÇÏ¿©µµ ³ôÀº ¼öÁØÀÇ Á¤È®µµ¸¦ ´Þ¼ºÇÒ ¼ö ÀÖÀ½À» ½ÇÇèÀ» ÅëÇØ º¸¿´´Ù.
¿µ¹®³»¿ë
(English Abstract)
The importance of intelligent command control systems for supporting commanders in their decision-making processes becomes increasingly recognized. In order to establish an intelligent command control system, knowledge bases need to be constructed, and accordingly, a hypothesis-generation method for AI-based battlefield analyses was proposed in a recent study. Since it is necessary to extract knowledge elements related to the commander¡¯s needs or queries to support decision-making, we need to identify the relevance between hypotheses and extract only related ones. However, it is difficult for general neural networks or machine learning models to identify the relevance between hypotheses because the knowledge elements within the hypotheses are interconnected to each other with a hierarchy. In this paper, we propose a contextual learning model based on hierarchical graph neural networks that effectively aggregates battlefield knowledge elements within given hypotheses, by utilizing the hierarchical structure of the hypotheses, and accurately identifies the relevance between the hypotheses. Through experiments, we demonstrate that our model is more accurate than the two baseline approaches, and especially, it achieves a high level of accuracy even when it uses only 40% of the data for training.
Å°¿öµå(Keyword) µ¥ÀÌÅÍ ¸¶ÀÌ´×   ÆÐÅÏ ¸¶ÀÌ´×   ÁÖ±âÀû ÆÐÅÏ ¸¶ÀÌ´×   À̵¿ °æ·Î µ¥ÀÌÅÍ   Data Mining   Pattern Mining   Periodic Pattern Mining   Trajectory Data   ÀüÀå»óȲ ºÐ¼®   ÀüÀå»óȲ ÀνĠ  ±×·¡ÇÁ ½Å°æ¸Á   ÀΰøÁö´É   Battlefield Analysis   Battlefield Awareness   Graph Neural Network   Artificial Intelligence  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå