• 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) ÁØÁöµµ ÀÌ»ó ŽÁö ¸ðµ¨ ±â¹Ý ½ÅÁ¾ À¯Çü »ç±â ŽÁö ¹æ¹ý
¿µ¹®Á¦¸ñ(English Title) Loan Fraud Detection Method based on Semi-supervised Anomaly Detection Models
ÀúÀÚ(Author) Á¶ÁØ¿µ   ÀåÀç¼®   ÀÓº¸¿µ   ±ÇÇõÀ±   Joonyoung Cho   Jaeseok Jang   Boyoung Lim   Hyuk-Yoon Kwon  
¿ø¹®¼ö·Ïó(Citation) VOL 38 NO. 03 PP. 0107 ~ 0120 (2022. 12)
Çѱ۳»¿ë
(Korean Abstract)
±ÝÀ¶±â¾÷¿¡ ´ëÇÑ ÀÌ¿ëÀÚÀÇ ÀǵµÀûÀÎ ´ëÃâ ÀÌÈÄ ¿¬Ã¼ÇÏ´Â À¯ÇüÀÇ ¿¬Ã¼ »ç±â´Â ±â¾÷¿¡ ¸·´ëÇÑ ¼ÕÇظ¦ ³¢Ä£´Ù. ƯÈ÷ Åë´ëȯ ¹× ½ÅÁ¾»ç±âÀÎ ¹«¿¬Ã¼È¸»ý, ÃÊȸÂ÷ ¿¬Ã¼´Â ½Å¿ëµµ¸¦ ¿Ö°î½ÃÄÑ Á¤»óÀûÀÎ ´ëÃâ ÀÌ¿ëÀڵ鿡 ´ëÇÑ ´ëÃâ °¡´É¼ºÀ» ³·Ãß°í ÅõÀÚÀڵ鿡°Ô ¸·´ëÇÑ ¼ÕÇظ¦ ³¢Ä¡Áö¸¸, ŽÁöµÈ Ƚ¼ö°¡ ±Ø¼Ò¼öÀÌ¸ç »ç±â µ¥ÀÌÅÍ °£¿¡ ÀûÀº À¯»ç¼ºÀ¸·Î ÀÎÇØ ±âÁ¸ÀÇ ¸Ó½Å·¯´× ¸ðµ¨·Î´Â ŽÁöÇϱ⠾î·Æ´Ù´Â ¹®Á¦°¡ ÀÖ´Ù. º» ¿¬±¸¿¡¼­´Â ŽÁöÇÏ´Â ´ë»óÀÌ Àüü¿¡¼­ 0.1% ¼öÁØÀÇ ±Ø¼Ò¼ö ½ÅÁ¾ À¯Çü »ç±â ŽÁö¸¦ ¸ñÇ¥·ÎÇÏ¿©, ¶óº§¸µµÈ »ç±â µ¥ÀÌÅÍ¿Í ¶óº§¸µµÇÁö ¾ÊÀº »ç±â µ¥ÀÌÅ͸¦ °°ÀÌ ÇнÀ¿¡ ÀÌ¿ëÇÏ´Â ÁØÁöµµ ÀÌ»ó ŽÁö ±â¹Ý »ç±â ŽÁö ¹æ¹ýÀ» ÃÖÃÊ·Î Á¦¾ÈÇÏ¿´´Ù. º» ¿¬±¸¿¡¼­´Â Á¤»ó »ùÇÃÀ» Á¤È®È÷ ºÐ·ùÇÏ´Â °Íº¸´Ù »ç±â¸¦ Á¤È®È÷ ŽÁöÇÏ´Â °Í¿¡ ´õ ÁßÁ¡À» µÎ´Â Æò°¡ ¹æ½ÄÀÎ Recall-BEP¸¦ Á¤ÀÇÇÏ¿© »ç¿ëÇÑ´Ù. ½ÇÇè °á°ú, ÁØÁöµµ ÀÌ»óÄ¡ ŽÁö ±â¹Ý ¹æ¹ý·ÐÀº ¶óº§¸µµÈ ÀÌ»óÄ¡ µ¥ÀÌÅ͸¦ È°¿ëÇÏ¿© ÇнÀÇÏ´Â ±âÁ¸ ¸ðµ¨¿¡ ºñÇØ Àü¹ÝÀûÀ¸·Î ÁÁÀº ¼º´ÉÀ» º¸¿´°í, ƯÈ÷ DevNet ±â¹Ý ¹æ¹ý·ÐÀº Recall-BEP¸¦ ±âÁØÀ¸·Î ±âÁ¸ ¸ðµ¨¿¡ ºñÇØ 17.39%±îÁö Çâ»óµÇ¾ú´Ù. À̸¦ ÅëÇØ ±Ø¼Ò¼öÀÇ ¶óº§¸µ µ¥ÀÌÅ͸¦ °¡Áö´Â ½ÅÁ¾ À¯Çü »ç±â¸¦ ŽÁöÇϱâ À§Çؼ­, ¶óº§¸µµÈ »ç±â »ùÇø¸À» »ç¿ëÇϱ⠺¸´Ù ¶óº§¸µµÇÁö ¾ÊÀº »ùÇÃÀ» Ãß°¡·Î È°¿ëÇÏ´Â ¹æ¹ý·ÐÀÇ È¿¿ë¼ºÀ» º¸¿´´Ù.
¿µ¹®³»¿ë
(English Abstract)
Delinquency fraud in which users intentionally overdue their loans causes enormous damage to financial companies. In particular, new types of frauds such as full repaying of existing loans, unpaid rehabilitation, and first-time delinquency distort creditworthiness reduce the execution of normal loans and cause enormous damages to investors. However, detecting them with the existing machine learning-based models is difficult due to very few fraud samples and low similarities between fraud samples. In this study, we focus on the fact that the target frauds are very few, which account only for 0.1% of the total datasets. Our study is the first research effort for a fraud detection methodology using semi-supervised anomaly detection that uses not only labeled fraud samples but also unlabeled potential fraud samples. In this study, we use Recall-BEP as an evaluation metric focusing more on accurately detecting frauds than on accurately classifying normal samples. Through the experiments, we show that the proposed semi-supervised based fraud detection methodology outperforms the existing comparison model based on the supervised learning. In particular, DevNet improves Recall-BEP by 17.39% compared to the comparison model. This result indicates the effectiveness of utilizing potential unlabeled fraud samples to detect new types of frauds than using only labeled fraud samples.
Å°¿öµå(Keyword) ´ëÃâ »ç±â   »ç±â ŽÁö   ÀÌ»óÄ¡ ŽÁö   ÁØÁöµµ ÇнÀ ¸ðµ¨   Loan fraud   Fraud detection   Anomaly detection   Semi-supervised models  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå