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

»çÀÌÆ®¸Ê

Loading..

Please wait....

±¹³» ³í¹®Áö

Ȩ Ȩ > ¿¬±¸¹®Çå > ±¹³» ³í¹®Áö > Çѱ¹Á¤º¸°úÇÐȸ ³í¹®Áö > Á¤º¸°úÇÐȸ³í¹®Áö (Journal of KIISE)

Á¤º¸°úÇÐȸ³í¹®Áö (Journal of KIISE)

Current Result Document : 7 / 40 ÀÌÀü°Ç ÀÌÀü°Ç   ´ÙÀ½°Ç ´ÙÀ½°Ç

ÇѱÛÁ¦¸ñ(Korean Title) Æ÷Áö¼Ç ÀÎÄÚµù ±â¹Ý S3-Net¸¦ ÀÌ¿ëÇÑ Çѱ¹¾î ±â°è µ¶ÇØ
¿µ¹®Á¦¸ñ(English Title) Korean Machine Reading Comprehension using S3-Net based on Position Encoding
ÀúÀÚ(Author) ¹ÚõÀ½   ÀÌâ±â   ±èÇö±â   Choeneum Park   Changki Lee   Hyunki Kim  
¿ø¹®¼ö·Ïó(Citation) VOL 46 NO. 03 PP. 0234 ~ 0240 (2019. 03)
Çѱ۳»¿ë
(Korean Abstract)
S3-NetÀº Simple Recurrent Unit (SRU)°ú ÀÚ±â ÀÚ½ÅÀÇ RNN sequence¿¡ ´ëÇÏ¿© ¾îÅÙ¼Ç °¡ÁßÄ¡(attention weight)¸¦ °è»êÇÏ´Â Self-Matching Networks¸¦ ±â¹ÝÀ¸·Î ±â°è µ¶ÇØ ÁúÀÇ ÀÀ´äÀ» ÇØ°áÇÏ´Â µö ·¯´× ¸ðµ¨ÀÌ´Ù. ±â°è µ¶ÇØ ÁúÀÇ ÀÀ´ä¿¡¼­ Áú¹®¿¡ ´ëÇÑ ´äÀº ¹®¸Æ ³»¿¡¼­ ¹ß»ýÇϴµ¥, ÇϳªÀÇ ¹®¸ÆÀº ¿©·¯ ¹®ÀåÀ¸·Î ÀÌ·ïÁö±â ¶§¹®¿¡ ÀÔ·Â ½ÃÄö½ºÀÇ ±æÀÌ°¡ ±æ¾îÁ® ¼º´ÉÀÌ ÀúÇϵǴ ¹®Á¦°¡ ÀÖ´Ù. º» ³í¹®¿¡¼­´Â ÀÌ¿Í °°ÀÌ ¹®¸ÆÀÌ ±æ¾îÁ® ¼º´ÉÀÌ ÀúÇϵǴ ¹®Á¦¸¦ ÇØ°áÇϱâ À§ÇÏ¿© ¹®Àå ´ÜÀ§ÀÇ ÀÎÄÚµùÀ» Ãß°¡ÇÑ °èÃþ¸ðµ¨°ú, ´Ü¾î ¼ø¼­ Á¤º¸¸¦ È®ÀÎÇÏ´Â Æ÷Áö¼Ç ÀÎÄÚµùÀ» Àû¿ëÇÑ S3-NetÀ» Á¦¾ÈÇÑ´Ù. ½ÇÇè °á°ú, º» ³í¹®¿¡¼­ Á¦¾ÈÇÑ S3-Net ¸ðµ¨ÀÌ Çѱ¹¾î ±â°è µ¶ÇØ µ¥ÀÌÅÍ ¼Â¿¡¼­ ±âÁ¸ÀÇ S2-Netº¸´Ù ¿ì¼öÇÑ(single test) EM 69.43%, F1 81.53%, (ensemble test) EM 71.28%, F1 82.67%ÀÇ ¼º´ÉÀ» º¸¿´´Ù.
¿µ¹®³»¿ë
(English Abstract)
S3-Net is a deep learning model that is used in machine reading comprehension question answering (MRQA) based on Simple Recurrent Unit and Self-Matching Networks that calculates attention weight for own RNN sequence. The answers to the questions in the MRQA occur within the passage, because any passage is made up of several sentences, so the length of the input sequence becomes longer and the performance deteriorates. In this paper, a hierarchical model that adds sentence-level encoding and S3-Net that applies position encoding to check word order information to solve the problem of long-term context degradation are proposed. The experimental results show that the S3-Net model proposed in this paper has a performance of 69.43% in EM and 81.53% in F1 for single test, and 71.28% in EM and 82.67 in F1 for ensemble test.
Å°¿öµå(Keyword) S3-Net   Çѱ¹¾î ±â°èµ¶ÇØ   Æ÷Áö¼Ç ÀÎÄÚµù   µö·¯´×   S3-Net   Korean machine reading comprehension   position encoding   deep learning  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå