Ç¥ÁØÈ­ Âü¿©¾È³»

TTAÀÇ Ç¥ÁØÇöȲ

Ȩ > Ç¥ÁØÈ­ °³¿ä > TTAÀÇ Ç¥ÁØÇöȲ

Ç¥ÁعøÈ£ TTAK.KO-11.0274 ±¸Ç¥ÁعøÈ£
Á¦°³Á¤ÀÏ 2020-12-10 ÃÑÆäÀÌÁö 23
ÇѱÛÇ¥Áظí ÀÓº£µðµå ½Ã½ºÅÛ¿¡¼­ ½Å°æ¸ÁÀÇ ÀúÀå ¹× ½ÇÇàÀ» À§ÇÑ ÄÁÅ×ÀÌ³Ê ¸ðµ¨
¿µ¹®Ç¥Áظí Container Model for Storing and Executing Neural Networks in Embedded Systems
Çѱ۳»¿ë¿ä¾à º» Ç¥ÁØÀº ÀÓº£µðµå ½Ã½ºÅÛ»ó¿¡¼­ ½Å°æ¸ÁÀÇ Ã߷п¡ ÇÊ¿äÇÑ ½Å°æ¸Á ¸ðµ¨ ÀÚü¿¡ ´ëÇÑ Á¤º¸ Ç¥Çö°ú ½Å°æ¸Á ¸ðµ¨ ½ÇÇà½Ã ÀÓº£µðµå ½Ã½ºÅÛ¿¡¼­ÀÇ Á¦¾à »çÇ×, ±â´É µî¿¡ ´ëÇÑ ºÎ°¡ Á¤º¸ Ç¥Çö°ú À̵é Á¤º¸¸¦ ÇϳªÀÇ µ¶¸³ÀûÀÎ ÆÄÀÏ È¤Àº Æú´õ¿¡ ÀúÀåÇÏ´Â ÄÁÅ×ÀÌ³Ê ±Ô°Ý¿¡ ´ëÇØ Á¤ÀÇÇÏ°í ÀÖ´Ù. º» Ç¥ÁØÀº ½Å°æ¸ÁÀÇ ±×·¡ÇÁ Á¤º¸¸¦ Ç¥ÇöÇÏ´Â ¹æ¹ý, ½Å°æ¸Á ³» ±×·¡ÇÁ¿¡¼­ »ç¿ëÇÏ´Â °¡ÁßÄ¡ µîÀÇ Á¤º¸¸¦ Ç¥ÇöÇÏ´Â ¹æ¹ý, ÇÊ¿äÇÑ ¿¬»ê ´É·Â, ¸Þ¸ð¸® ¿ä±¸·®, ÀÓº£µðµå ½Ã½ºÅÛÀÇ ÀÚ¿ø Á¦¾à »çÇ×, ÁÖ¿ä ±â´É µî ½Å°æ¸Á ½ÇÇà ½Ã¿¡ µµ¿òÀÌ µÇ´Â ºÎ°¡ Á¤º¸ Ç¥Çö ¹æ¹ý°ú À̵é Á¤º¸¸¦ ÇϳªÀÇ ÄÁÅ×À̳Ê(ÆÄÀÏ È¤Àº Æú´õ)·Î ÀúÀåÇÏ´Â ¹æ¹ý¿¡ ´ëÇØ Á¤ÀÇÇÏ°í ÀÖ´Ù.
¿µ¹®³»¿ë¿ä¾à This standard expresses information about the neural network model itself necessary for inference of neural networks on embedded systems, and expresses additional information about constraints and functions in the embedded systems when executing the neural network model, and stores these information in an independent file or folder. It defines the storage container standard. This standard is helpful when performing tasks such as a method of expressing graph information of a neural network, a method of expressing information such as weights used in graphs in a neural network, required computing power, memory requirements, resource constraints of embedded systems, and major functions. It defines the method of expressing the additional information and how to store the information in a single container (file or folder).
±¹Á¦Ç¥ÁØ
°ü·ÃÆÄÀÏ TTAK.KO-11.0274.pdf TTAK.KO-11.0274.pdf            

ÀÌÀü
ÀÓº£µðµå ½Ã½ºÅÛ¿¡¼­ ½Å°æ¸Á »óÈ£ ȣȯÀ» À§ÇÑ ÇнÀ µ¥ÀÌÅÍ ±¸Á¶ÀÇ ¿ä±¸»çÇ×
´ÙÀ½
ÀÓº£µðµå ½Ã½ºÅÛ¿¡¼­ÀÇ ºñÀü ÀÀ¿ë ÇÁ·Î±×·¥À» À§ÇÑ À¥ ±â¹Ý ¿ø°Ý ½Ã°¢È­ ÇÁ·ÎÅäÄÝ