Ç¥ÁØÈ­ Âü¿©¾È³»

TTAÀÇ Ç¥ÁØÇöȲ

Ȩ > Ç¥ÁØÈ­ °³¿ä > TTAÀÇ Ç¥ÁØÇöȲ

Ç¥ÁعøÈ£ TTAE.OT-10.0471 ±¸Ç¥ÁعøÈ£
Á¦°³Á¤ÀÏ 2025-12-05 ÃÑÆäÀÌÁö 317
ÇѱÛÇ¥Áظí À¥ ½Å°æ¸Á API
¿µ¹®Ç¥Áظí Web Neural Network API
Çѱ۳»¿ë¿ä¾à ÀÌ Ç¥ÁØÀº À¥ ºê¶ó¿ìÀú ȯ°æ¿¡¼­ ½Å°æ¸Á ±â¹Ý ¸Ó½Å·¯´× ¸ðµ¨ÀÇ Ãß·Ð(inference)À» °í¼º´ÉÀ¸·Î ½ÇÇàÇÒ ¼ö ÀÖµµ·Ï Áö¿øÇÏ´Â W3C Ç¥ÁØ APIÀÌ´Ù. ÀÌ API´Â CPU, GPU, NPU µî ´Ù¾çÇÑ Çϵå¿þ¾î °¡¼Ó±â¸¦ Ȱ¿ëÇØ À̹ÌÁö ºÐ·ù, À½¼º ÀνÄ, ÀÚ¿¬¾î ó¸® µî°ú °°Àº º¹ÀâÇÑ AI ÀÛ¾÷À» ½Ç½Ã°£À¸·Î À¥ ¾ÖÇø®ÄÉÀ̼ǿ¡¼­ ½ÇÇàÇÒ ¼ö ÀÖ°Ô ÇÑ´Ù. ¸ðµ¨Àº ¿¬»êÀÚ ±â¹ÝÀÇ °è»ê ±×·¡ÇÁ ÇüÅ·Π±¸¼ºµÈ´Ù. ¸ðµç ¿¬»êÀº Ŭ¶óÀÌ¾ðÆ® Ãø¿¡¼­ ¼öÇàµÇ¾î ¼­¹ö¿ÍÀÇ ºÒÇÊ¿äÇÑ µ¥ÀÌÅÍ Àü¼Û ¾øÀÌ °³ÀÎÁ¤º¸ º¸È£¿Í Áö¿¬ ½Ã°£ °¨¼Ò¶ó´Â ÀåÁ¡À» Á¦°øÇÑ´Ù. WebNN API´Â TensorFlow Lite, ONNX µî ±âÁ¸ ¸ðµ¨ Æ÷¸Ë°úµµ ¿¬µ¿ °¡´ÉÇÏ¿© À¥¿¡¼­ AI ±â´ÉÀ» ¼Õ½±°Ô ±¸ÇöÇÒ ¼ö ÀÖ´Â ±â¹ÝÀ» ¸¶·ÃÇÑ´Ù.
¿µ¹®³»¿ë¿ä¾à This standard is a W3C API designed to support high-performance execution of neural network-based machine learning model inference in web browser environments. The API leverages various hardware accelerators such as CPUs, GPUs, and NPUs to enable real-time execution of complex AI tasks like image classification, speech recognition, and natural language processing within web applications. Models are structured as operator-based computational graphs. All operations are performed on the client side, providing advantages such as user privacy protection and reduced latency by avoiding unnecessary data transmission to servers. The WebNN API also supports interoperability with existing model formats like TensorFlow Lite and ONNX, establishing a foundation for easy implementation of AI functionality on the web.
±¹Á¦Ç¥ÁØ
°ü·ÃÆÄÀÏ TTAE.OT-10.0471.pdf TTAE.OT-10.0471.pdf            

ÀÌÀü
´ÙÀ½