ÀÚ·á°Ë»ö-Ç¥ÁØ

Ȩ > ÀڷḶ´ç > ÀÚ·á°Ë»ö > Ç¥ÁØ

ÀÚ·á °Ë»ö°á°ú

°Ë»öÆäÀÌÁö·Î
Ç¥ÁØÁ¾·ù Á¤º¸Åë½Å±â¼úº¸°í¼­(TTAR)
Ç¥ÁعøÈ£ TTAR-10.0137 ±¸ Ç¥ÁعøÈ£
Á¦°³Á¤ÀÏ 2020-10-21 ÃÑ ÆäÀÌÁö 0
ÇÑ±Û Ç¥Áظí ÇÕ¼º°ö ½Å°æ¸Á ±âº» ¿¬»ê ÀÎÅÍÆäÀ̽º ±¸Çö »ç·Ê(±â¼úº¸°í¼­)
¿µ¹® Ç¥Áظí Convolutional Neural Network Basic Computational Interface Implementation Case(Technical Report)
ÇÑ±Û ³»¿ë¿ä¾à ÀÌ ±â¼úº¸°í¼­´Â ÀÓº£µðµå ½Ã½ºÅÛ»ó¿¡¼­ °æ·® µö·¯´×À» À§ÇÑ ÇÕ¼º°ö ½Å°æ¸Á ±âº» ¿¬»ê ÀÎÅÍÆäÀ̽º ±¸¼º°ú µ¿ÀÛ ±¸Á¶¸¦ Á¦½ÃÇÑ´Ù. º» ±â¼úº¸°í¼­¿¡¼­ Á¦½ÃÇÏ´Â ÇÕ¼º°ö ½Å°æ¸Á ±âº» ¿¬»ê ÀÎÅÍÆäÀ̽º´Â °¢°¢ÀÇ Çϵå¿þ¾î¿¡ ÃÖÀûÈ­µÈ ´Ù¾çÇÑ ±âÁ¸ ¶óÀ̺귯¸®µéÀ» Ç¥ÁØÈ­µÈ ÀÎÅÍÆäÀ̽º¸¦ ÅëÇØ °ü¸®Çϸç, CPU¿Í GPU ±â¹Ý ¿¬»êÀ» ¸ðµÎ Áö¿øÇÑ´Ù. ƯÈ÷ º» ±â¼úº¸°í¼­¿¡¼­ Á¦½ÃÇÑ ÇÕ¼º°ö ½Å°æ¸Á ±âº» ¿¬»ê ÀÎÅÍÆäÀ̽º´Â ´Ù¾çÇÑ ÀÓº£µðµå ½Ã½ºÅÛ¿¡¼­ GPU ±â¹Ý ¿¬»êÀ» Áö¿øÇϱâ À§ÇØ OpenCL ±â¹Ý µ¥ÀÌÅÍ ±¸Á¶¸¦ Áö¿øÇÑ´Ù. º» ±â¼úº¸°í¼­¿¡¼­ Á¦½ÃÇÑ ÇÕ¼º°ö ½Å°æ¸Á ±âº» ¿¬»ê ÀÎÅÍÆäÀ̽º ÂüÁ¶±¸ÇöÀº C++ ¾ð¾î Çü½ÄÀ¸·Î ±â¼úÇÏ¿´À¸³ª ´Ù¸¥ ¾ð¾î¿¡µµ Àû¿ëÀÌ °¡´ÉÇÏ´Ù.
¿µ¹® ³»¿ë¿ä¾à This technical report presents the CNN basic computational interface configuration and operation structure of a convolutional neural network in embedded systems. The CNN basic computationa interface presented in this technical report manages various CNN basic computational libraries through a standardized interface and supports both CPU and GPU-based operations. In particular, the CNN basic computational interface presented in this technical report supports OpenCL-based data structures to support GPU-based operations in various embedded systems. Although the CNN basic computational interface reference implementation presented in this technical report is written in C ++ language format, it can be applied to other languages.
°ü·Ã IPR È®¾à¼­ Á¢¼öµÈ IPR È®¾à¼­ ¾øÀ½
°ü·ÃÆÄÀÏ    TTAR-10.0137.pdf TTAR-10.0137.pdf
Ç¥ÁØÀÌ·Â
Ç¥Áظí Ç¥ÁعøÈ£ Á¦°³Á¤ÀÏ ±¸ºÐ À¯È¿
¿©ºÎ
IPR
È®¾à¼­
ÆÄÀÏ
ÇÕ¼º°ö ½Å°æ¸Á ±âº» ¿¬»ê ÀÎÅÍÆäÀ̽º ±¸Çö »ç·Ê(±â¼úº¸°í¼­) TTAR-10.0137 2020-10-21 Á¦Á¤ À¯È¿ ¾øÀ½ TTAR-10.0137.pdf