• 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 : 281 / 281

ÇѱÛÁ¦¸ñ(Korean Title) X-MOCNN : ÀÏ°ü¼º ³ôÀº ¼ÒÇÁÆ®¿þ¾î °áÇÔ ºÐ¼®À» À§ÇÑ ´ÙÁß Ãâ·Â ÇÕ¼º°ö ½Å°æ¸Á
¿µ¹®Á¦¸ñ(English Title) X-MOCNN : A Multi-Output Convolutional Neural Network for Consistent Software Defect Analysis
ÀúÀÚ(Author) ÀÌ´ÙÀΠ  ¿Àº´µÎ   ÃÖÇü   ±èÀ¯¼·   Da-In Lee   Byoung-Doo Oh   Hyung Choi   Yu-Seop Kim   ÀÌÀç¿í   ÃÖÁö¿ø   ·ù´ö»ê   ±è¼øÅ   Jaewook Lee   Jiwon Choi   Duksan Ryu   Suntae Kim  
¿ø¹®¼ö·Ïó(Citation) VOL 28 NO. 02 PP. 0096 ~ 0102 (2022. 02)
Çѱ۳»¿ë
(Korean Abstract)
¼ÒÇÁÆ®¿þ¾î °áÇÔ ºÐ¼®Àº ¼ÒÇÁÆ®¿þ¾îÀÇ Ç°ÁúÀ» º¸ÁõÇÏ´Â µ¥¿¡ Áß¿äÇÑ ¿¬±¸ ºÐ¾ß Áß ÇϳªÀÌ´Ù. °áÇÔ ºÐ¼®Àº ¿¹ÃøÇÏ°íÀÚ ÇÏ´Â Ãâ·Â¿¡ µû¶ó °áÇÔ ¿¹Ãø, °áÇÔ È¸±Í, °áÇÔ ½É°¢µµ ¿¹Ãø µîÀ¸·Î ºÐ·ùµÈ´Ù. °¢ Ãâ·Â¿¡¼­ ¶Ù¾î³­ ¿¹Ãø¼º´ÉÀ» º¸À̱â À§ÇØ ¸¹Àº ¿¬±¸°¡ Á¦¾ÈµÇ¾ú´Ù. ÇÏÁö¸¸, ½ÇÁ¦ »ç¿ëÀÚ°¡ ¿©·¯ ¸ðµ¨·Î ¿©·¯ Ãâ·Â Á¤º¸¸¦ ¾ò°íÀÚ ÇÏ´Â °æ¿ì, ¸ðµ¨µéÀÇ Ãâ·Â °£ ÀÏ°ü¼ºÀÌ ³·Àº ¹®Á¦°¡ Á¸ÀçÇÑ´Ù. º» ³í¹®¿¡¼­´Â °áÇÔ・ºñ°áÇÔ, ¹ö±×ÀÇ ¼ö, ¹ö±×ÀÇ ½É°¢µµ 3°¡Áö Ãâ·ÂÀ» µ¿½Ã¿¡ ÇϳªÀÇ ¸ðµ¨·Î ¿¹ÃøÇØ Ãâ·Â °£ ÀÏ°ü¼ºÀ» ³ôÀÌ°íÀÚ ÇÑ´Ù. À̸¦ À§ÇØ XGBoost¿Í ´ÙÁß Ãâ·Â 1Â÷¿ø ÇÕ¼º°ö ½Å°æ¸ÁÀ» °áÇÕÇÑ ¸ðµ¨ÀÎ X-MOCNN(XGBoost with Multi-Output 1-d Convolutional Neural Network)À» Á¦¾ÈÇÑ´Ù. ÀÌ ¸ðµ¨ÀÇ È¿¿ë¼ºÀ» °ËÁõÇϱâ À§ÇØ ´ÜÀÏ Ãâ·Â ½Å°æ¸Á ¸ðµ¨, ¸Ó½Å ·¯´× ¸ðµ¨°ú ¼º´ÉÀ» ºñ±³ÇÏ¿´´Ù. ±× °á°ú X-MOCNNÀº ±âÁ¸ÀÇ ´ÜÀÏ Ãâ·Â ½Å°æ¸Á ¸ðµ¨ ´ëºñ ³ôÀº ¼º´ÉÀ» º¸¿´À¸¸ç °¢ ¿¹Ãø °á°ú °£ ÀÏ°ü¼º Á¤µµµµ ¶Ù¾î³µ´Ù. À̸¦ ÅëÇØ ¼ÒÇÁÆ®¿þ¾î ÇÁ·ÎÁ§Æ®¿¡¼­ ÇϳªÀÇ ¸ðµ¨·Î ¿©·¯ Ãâ·ÂÀ» ¿¹ÃøÇØ Ç°Áúº¸Áõ ÀÚ¿øÀ» È¿°úÀûÀ¸·Î ÇÒ´çÇÒ ¼ö ÀÖ´Ù.
¿µ¹®³»¿ë
(English Abstract)
Software defect analysis is one of the most important research fields in software engineering to ensure software quality. According to the predicted output, defect analysis is classified into defect prediction, defect regression, and defect-severity prediction. Many studies have been proposed to outperform at each output. However, when an actual user wants to obtain multiple output information with multiple models, there is a problem of low consistency between the outputs of the models. In this work, we simultaneously predict the following three outputs with one model: bug identification, the number of bugs, and the severity of bugs to increase consistency between outputs. To this end, we propose an X-MOCNN, which combines XGBoost and a Multi-Output 1-d Convolutional Neural Network. We compare the performance of a single-output neural network model and a machine learning model to verify the effectiveness of our proposed model. As a result of the experiment, X-MOCNN shows higher performance and consistency between outputs than the other models. Therefore, it is expected that the method presented by us can effectively allocate quality assurance resources by predicting multiple outputs with one model in a software project.
Å°¿öµå(Keyword) PSC ¹Ú½º °Å´õ±³   Impact-Echo ½ÅÈ£   LSTM ¿ÀÅä ÀÎÄÚ´õ   ÀáÀç º¤ÅÍ   ÀÌ»ó ŽÁö   PSC box girder bridge   impact-echo signal   LSTM auto-encoder   latent vector   anomaly Detection   ¼ÒÇÁÆ®¿þ¾î °áÇÔ ºÐ¼®   ´ÙÁß Ãâ·Â ÇнÀ   ´ÙÁß Ãâ·Â ½Å°æ¸Á ¸ðµ¨   ÇÕ¼º°ö ½Å°æ¸Á   ½ÉÃþ ÇнÀ   software defect analysis   multi-output learning   multi-output neural network   convolutional neural network   deep learning  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå