Á¤º¸°úÇÐȸ ³í¹®Áö B : ¼ÒÇÁÆ®¿þ¾î ¹× ÀÀ¿ë
ÇѱÛÁ¦¸ñ(Korean Title) |
½º¸¶Æ®Æù ¸ÖƼ¸ð´Þ ¼¾¼ ±â¹Ý °³ÀÎÈ ÇàÀ§¸ðµ¨¸µ ¹× ½Ç½Ã°£ ÇàÀ§ÀÎÁö |
¿µ¹®Á¦¸ñ(English Title) |
Personalized Activity Modeling and Real-time Activity Recognition based on Smartphone Multimodal Sensors |
ÀúÀÚ(Author) |
ÇѸ¸Çü
À̽·æ
Manhyung Han
Sungyoung Lee
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 40 NO. 06 PP. 0332 ~ 0341 (2013. 06) |
Çѱ۳»¿ë (Korean Abstract) |
½º¸¶Æ®Æù°ú °°Àº ¸ð¹ÙÀÏ ±â±âÀÇ ¹ßÀüÀ¸·Î ÀÎÇØ ´Ù¾çÇÑ ¼¾¼¸¦ ÅëÇØ »ç¿ëÀÚÀÇ Àǵµ³ª ¿ä±¸»çÇ×À» ÀÎÁöÇÏ°íÀÚ ÇÏ´Â ÇàÀ§ÀÎÁö°¡ È°¹ßÈ÷ ¿¬±¸µÇ°í ÀÖ´Ù. ±âÁ¸ÀÇ ÇàÀ§ÀÎÁö ±â¼úÀº ÇàÀ§ µ¥ÀÌÅÍÀÇ ¼öÁý°ú 󸮰¡ ºÐ¸®µÇ¾î ÀÖ¾î ¸ÖƼ¸ð´Þ ¼¾¼·ÎºÎÅÍ ¼öÁýÇÑ ¹æ´ëÇÑ ¾çÀÇ µ¥ÀÌÅ͸¦ ¿ÀÇÁ¶óÀο¡¼ ¸ðµ¨¸µ°ú ÀÎÁö°¡ ¼öÇàµÇ¹Ç·Î ƯÁ¤ »ç¿ëÀÚ¿¡ °³ÀÎÈµÈ ÇàÀ§ÀÎÁö°¡ ¾î·Æ°í, »ç¿ëÀÚ°¡ Á÷Á¢ ÀÚ½ÅÀÇ Æ¯Á¤ÇàÀ§¸¦ Ãß°¡Çϰųª ½º½º·Î ÇàÀ§¸ðµ¨À» ¸¸µé ¼ö ÀÖ´Â ÇÁ·¹ÀÓ¿öÅ©ÀÇ ºÎÀç·Î °³ÀÎÈµÈ ¶óÀÌÇÁ·Î±×ÀÇ ¼öÁýÀÌ ¾î·Æ´Ù´Â ÇÑ°è°¡ ÀÖ´Ù. º» ³í¹®¿¡¼´Â ½º¸¶Æ®Æù¿¡¼ °³ÀÎÈµÈ ÇàÀ§ ¸ðµ¨¸µ ¹× ½Ç½Ã°£ ÇàÀ§ÀÎÁö¸¦ À§ÇØ, Naive Bayes ¾Ë°í¸®ÁòÀ» È®ÀåÇÑ ÀûÀÀÇü Naive Bayes(A-NB) ¾Ë°í¸®Áò°ú À̸¦ ±â¹ÝÀ¸·Î ÇÑ °èÃþÀû ÇàÀ§ÀÎÁö ÇÁ·¹ÀÓ¿öÅ©(HARF)¸¦ Á¦¾ÈÇÑ´Ù. À̸¦ ÅëÇØ ½º¸¶Æ®Æù ȯ°æ¿¡¼ »ç¿ëÀÚ°¡ ½º½º·Î ÀÚ½ÅÀÇ ÇàÀ§¸¦ ¸ðµ¨¸µÇϰųª Ãß°¡ÇÒ ¼ö ÀÖÀ¸¸ç, Naive Bayes¿¡ ºñÇØ ³ôÀº Á¤È®µµ¿Í ¸ð¹ÙÀÏ È¯°æ¿¡¼ ½Ç½Ã°£ ÇàÀ§ÀÎÁö°¡ °¡´ÉÇÏ´Ù. Á¦¾È ¾Ë°í¸®ÁòÀÇ Æò°¡¸¦ À§ÇØ ½º¸¶Æ®Æù ¾îÇø®ÄÉÀ̼ÇÀ» °³¹ßÇÏ¿© 15°³ÀÇ ÇàÀ§¸¦ ½ÇÇèÇÏ¿´À¸¸ç Æò±Õ 92.96%ÀÇ ³ôÀº Á¤È®µµ¸¦ º¸¿´´Ù. |
¿µ¹®³»¿ë (English Abstract) |
Activity recognition for the purposes of recognizing a user¡¯s intentions using multimodal sensors is becoming a widely researched topic largely based on the prevalence of the smartphone. Previous studies have reported the difficulty in recognizing personalized activities of individual users given that the collection and processing of the vast amount of activity data from multimodal sensors are separated and performed on off-line. In addition, recognizing personalized life-logs is difficult due to the absence of a framework which enables the addition of activities by the user themselves. In this paper, we propose an adaptive Naive Bayes (A-NB) algorithm and hierarchical activity recognition framework (HARF) which extends the Naive Bayes approach in an effort to personalizes the process of activity modeling & real-time activity recognition. Based on this approach, the users can add or model their own activities by themselves with a smartphone. The proposed algorithm demonstrates relatively higher accuracy than the Naive Bayes approach and also enables the recognition of the user¡¯s activities in a mobile environment. For the purposes of evaluation, we have developed a smartphone application. Based on this platform, the experimental results demonstrate that the proposed algorithm has the ability to classify fifteen activities with an average accuracy of 92.96%. |
Å°¿öµå(Keyword) |
ÇàÀ§ÀÎÁö
½º¸¶Æ®Æù
¸ÖƼ¸ð´Þ¼¾¼
Naive Bayes
¶óÀÌÇÁ·Î±×
°³ÀÎÈ
activity recognition
smartphone
multimodal sensors
life-log
personalization
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|