• 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 : 9 / 12 ÀÌÀü°Ç ÀÌÀü°Ç   ´ÙÀ½°Ç ´ÙÀ½°Ç

ÇѱÛÁ¦¸ñ(Korean Title) CNNÀ» È°¿ëÇÑ IoT ½ºÆ®¸² µ¥ÀÌÅÍ ÆÐÅÏ ºÐ·ù ±â¹ý
¿µ¹®Á¦¸ñ(English Title) Pattern Classification for IoT Stream Data using Convolutional Neural Networks
ÀúÀÚ(Author) ±è°æÁÖ   ¿À¼Ò¿¬   À̹μö   Kyeongjoo Kim   Soyeon Oh   Minsoo Lee  
¿ø¹®¼ö·Ïó(Citation) VOL 35 NO. 02 PP. 0106 ~ 0115 (2019. 08)
Çѱ۳»¿ë
(Korean Abstract)
»ç¹° ÀÎÅͳÝ(Internet of Things, IoT) ȯ°æÀÇ ¹ß´Þ·Î ´Ù¾çÇÑ Á¾·ùÀÇ ¼¾¼­µé·ÎºÎÅÍ ´ë·®ÀÇ µ¥ÀÌÅÍ°¡ »ý¼º µÇ°í ÀÖÀ¸¸ç, À̸¦ ¼öÁý, °ü¸® ¹× ºÐ¼®Çϱâ À§ÇÑ ºòµ¥ÀÌÅÍ ±â¼úÀÌ Áß¿äÇØÁö°í ÀÖ´Ù. ÃÖ±Ù¿¡´Â ½Ç½Ã°£À¸·Î »ý¼ºµÇ´Â ´ë¿ë·®ÀÇ IoT µ¥ÀÌÅÍ ºÐ¼®¿¡ µö·¯´× ±â¼úÀ» È°¿ëÇÏ¿© ƯÁ¤ µ¥ÀÌÅÍ ÆÐÅÏÀ̳ª °æÇ⼺ÀÇ ºÐ¼®À» ¼öÇàÇϱâ À§ÇÑ ¿¬±¸°¡ ÁøÇàµÇ°í ÀÖ´Ù. º» ³í¹®¿¡¼­´Â ÇコÄÉ¾î µî IoT ±â¹Ý ¼­ºñ½º¿¡ÀÇ È°¿ë °¡´É¼ºÀÌ ³ôÀº ½ºÆ®¸²µ¥ÀÌÅÍ Áß ÇϳªÀÎ ECG(Electrocardiogram, ½ÉÀüµµ) µ¥ÀÌÅÍ¿¡ ´ëÇÏ¿©, µö·¯´× ¸ðµ¨À» ¼³°è ¹× Àû¿ëÇÔÀ¸·Î½á È¿À²ÀûÀÎ ºÐ¼®À» °¡´ÉÇϵµ·Ï ÇÏ¿´´Ù. ¸ÕÀú, ½ºÆ®¸² ECG µ¥ÀÌÅÍÀÇ ÆÐÅÏ ºÐ·ù¸¦ À§ÇÏ¿© ÇÕ¼º°ö ½Å°æ¸Á (Convolutional Neural Networks, CNN) ±â¹ÝÀÇ µö·¯´× ¸ðµ¨À» ¼³°èÇÏ°í, À̸¦ ÃÖÀûÈ­Çϱâ À§ÇÑ ´Ù¾çÇÑ ÆĶó¹ÌÅ͵éÀ» °¢°¢ ¸ðµ¨ÀÇ ±¸Á¶¿Í ÇнÀ¿¡ °ü·ÃÇÑ ÆĶó¹ÌÅ͵é·Î ºÐ¸®ÇÏ¿© ½ÇÇèÀ» ¼³°è ¹× ÁøÇàÇÏ¿´´Ù. ¶ÇÇÑ, ºÐ·ù ÀÛ¾÷ÀÇ Ãß°¡ÀûÀÎ ¼º´É Çâ»óÀ» À§ÇÏ¿© ECG ½ºÆ®¸² µ¥ÀÌÅÍ¿¡ ´ëÇÑ Àüó¸® ±â¹ýÀ» °í¾ÈÇÏ¿© Àû¿ëÇØ º¸¾Ò´Ù. ÀÌ·¯ÇÑ ´Ù¾çÇÑ Á¶°ÇÀ» ±â¹ÝÀ¸·Î ¼³°èµÈ ½ÇÇèµéÀº ¼­·Î ´Ù¸¥ ¼¾¼­¿¡¼­ ¼­·Î ´Ù¸¥ ¸ñÀûÀ¸·Î ¼öÁýµÇ¾î ¼­·Î ´Ù¸¥ Ư¼ºÀ» °®´Â µÎ °¡ÁöÀÇ ECG ½ºÆ®¸² µ¥ÀÌÅÍ ¼¼Æ®¿¡ ´ëÇÏ¿© °¢°¢ ¼öÇàµÇ¾ú´Ù. ±× °á°ú, ·¹À̾ ±íÀ»¼ö·Ï ¹èÄ¡ Å©±â°¡ Å« ÇнÀ ¸ðµ¨Àϼö·Ï IoT ½ºÆ®¸² µ¥ÀÌÅÍÀÇ ÆÐÅÏ ºÐ·ù¿¡ ¿ëÀÌÇÑ ¸ðµ¨ ±¸Á¶¶ó´Â °á·ÐÀ» ¾òÀ»¼ö ÀÖ¾ú´Ù.
¿µ¹®³»¿ë
(English Abstract)
These days due to the development of the Internet of Things environment, big data technology is becoming important for collecting and managing large amounts of data. Recent studies are being conducted to incorporate deep learning technology into Internet of Things(IoT) data analysis in order to classify the specific pattern and trends. In this paper, ECG(Electrocardiogram) data, which could be useful for IoT services, is the input steam data, and a deep learning model structure suitable for data characteristics is found, so that IoT data analysis is efficiently performed. In order to classify the IoT stream data pattern, the experiments were conducted to find the best suitable model structure using the convolutional neural networks. To optimize the CNN, various models and parameter values were used to design various experiments. Also to enhance the classification performance, a preprocessing step is added to the existing convolutional neural networks model. The model structure parameters and the model learning parameters are divided into two major conditions. The experiment environment is set up and applied to two time series data with different characteristics. It is concluded that the deeper the layer and the larger the batch size, the easier model structure for IoT data pattern classification.
Å°¿öµå(Keyword) IoT   stream data   deep learning   pattern analysis   ½ºÆ®¸² µ¥ÀÌÅÍ   µö·¯´× ÆÐÅÏ ºÐ·ù  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå