• 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¸°úÇÐȸ ÄÄÇ»ÆÃÀÇ ½ÇÁ¦ ³í¹®Áö (KIISE Transactions on Computing Practices)

Á¤º¸°úÇÐȸ ÄÄÇ»ÆÃÀÇ ½ÇÁ¦ ³í¹®Áö (KIISE Transactions on Computing Practices)

Current Result Document : 7 / 10 ÀÌÀü°Ç ÀÌÀü°Ç   ´ÙÀ½°Ç ´ÙÀ½°Ç

ÇѱÛÁ¦¸ñ(Korean Title) ¼Ò·® µ¥ÀÌÅÍ ¼Â ±â¹Ý Çǵå¹é Áöµµ ÇнÀ
¿µ¹®Á¦¸ñ(English Title) Feedback-supervised Learning with A Small Dataset
ÀúÀÚ(Author) ±èÁøÈñ   ÀÌ¿õÈñ   ±è¿µÈÆ   Jinhee Kim   Woonghee Lee   Younghoon Kim  
¿ø¹®¼ö·Ïó(Citation) VOL 25 NO. 02 PP. 0130 ~ 0135 (2019. 02)
Çѱ۳»¿ë
(Korean Abstract)
ÃÖ±Ù ÀÚµ¿ ºÐ·ùÈ­¿¡¼­ µö·¯´×ÀÌ °ý¸ñÇÒ¸¸ÇÑ ¼º´ÉÇâ»óÀ» º¸ÀÌ°í ÀÖ´Ù. µö·¯´×Àº ¸¹Àº µ¥ÀÌÅ;çÀÌ ÁÖ¾îÁ³À» ¶§ ³ôÀº ¼º´ÉÀÌ º¸ÀåµÇ´Âµ¥, ¸¸¾à Ãʱâ ÇнÀ µ¥ÀÌÅ;çÀÌ ÃæºÐÇÏÁö ¾ÊÀ» ¶§´Â °úÇнÀÀ¸·Î ÀÎ ÇØ ¼º´ÉÀÌ ÀúÇϵǴ ¹®Á¦°¡ ÀÖ´Ù. ÇÏÁö¸¸ Çö½Ç ¼¼°è¿¡¼­ ¸¹Àº µ¥ÀÌÅ͸¦ ¼öÁýÇÏ´Â µ¥´Â ³ôÀº ºñ¿ë°ú ½Ã°£ ÀÌ ¼Ò¿äµÈ´Ù. ¶ÇÇÑ °¡¿ëµ¥ÀÌÅÍ´Â ¼Ò·®À̳ª ºÐ·ù±âÁØÀ» ÁöµµÇØÁÙ ¼ö ÀÖ´Â ºÐ¾ß Àü¹®°¡°¡ ÀÖ´õ¶óµµ »ç¶÷ÀÇ Çǵå¹éÀ» ÇнÀ ¼º´É Çâ»ó¿¡ Á÷Á¢ È°¿ëÇÒ ¼ö ÀÖ´Â ¹æ¹ýÀÌ ¾ø´Ù. º» ¿¬±¸¿¡¼­´Â µö·¯´× ¸ðµ¨À» »ç¿ëÇßÀ» ¶§, ºÐ¾ß Àü¹®°¡ÀÇ Çǵå¹éÀ» ÅëÇØ ¼Ò·® µ¥ÀÌÅÍÀÇ ºÐ·ù ¼º´ÉÀ» Çâ»ó½ÃÅ°°íÀÚ ÇÑ´Ù. À̸¦ À§ÇØ ¾ÆÀÌÅÛ À̹ÌÁö¸¸ À» »ç¿ëÇÑ ºÐ·ù ¼º´É°ú ºÐ¾ß Àü¹®°¡ÀÇ Çǵå¹é °úÁ¤À» ÇÔ²² ÇнÀÇÑ ¾ÆÀÌÅÛ À̹ÌÁö ºÐ·ù ¼º´ÉÀ» ºñ±³ÇÏ¿´´Ù. ½ÇÇèÀ» ÅëÇØ ºÐ¾ß Àü¹®°¡ÀÇ Çǵå¹éÀ» ÇнÀÇÏ¿© À̹ÌÁö¸¦ ºÐ·ùÇÒ °æ¿ì Çǵå¹éÀÌ ¾ø´Â °æ¿ìº¸´Ù Á¤È®µµ°¡ Çâ»óµÊÀ» º¸¿´´Ù.
¿µ¹®³»¿ë
(English Abstract)
Lately, deep learning has had a significant performance improvement in automatic classification. Given a large set of data, deep learning guarantees high performance. When there is insufficient initial learning data, performance degrades due to the scientific method. However, it¡¯s practically expensive and time consuming to collect large sets of data. Additionally, there is no avenue to directly use human feedback to improve learning the experience, even if there is an expert in the field that can teach small bits on classification criteria of available data. In this paper, we use the deep learning model to improve the classification performance of small amount of data through field experts. For this study, we compared the classification performance using item image only to item image classification performance learned together with expert feedback process. Experiments have shown that the accuracy of classification of images by learning feedback from field experts is better than the one without feedback.
Å°¿öµå(Keyword) À̹ÌÁöºÐ·ù   Çǵå¹éÁöµµ   µö·¯´×   µ¥ÀÌÅͺÎÁ·¹®Á¦   °úÇнÀ¹®Á¦   image classification   feedback-supervised   deep learning   insufficient data   overfit  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå