2020.03.25 æè¡è¨äº DRIVE CHARTã«ããã AIæè¡ã®äºæ¥å¿ç¨ ã¼ã¢ãã«éçºãããµã¼ãã¹ãããã¤ã¾ã§ã¼ Part1 by Hidenori-Kuribayashi #automotive #AI #dena-techcon AIã·ã¹ãã é¨ã®æ æã¨ç³ãã¾ããAIç 究éçºã°ã«ã¼ãã®ããã¼ã¸ã£ã¼ã¨ãã¦ã主ã«ãDRIVE CHARTãã§ä½¿ç¨ããèªèæè¡ã®éçºãæ å½ãã¦ããã¾ããæã ã¯ãAIæè¡ãå¹çããå®ãµã¼ãã¹ã«é©ç¨ããããè¦ç´ æè¡éçºãããããã¯ãã¸ã®å®è£ ãä¸è²«ãã¦è¡ã£ã¦ãã¾ãã ãã®Blogè¨äºã¯ãDRIVE CHARTã«ããã AIæè¡ã®äºæ¥å¿ç¨ ã¼ã¢ãã«éçºãããµã¼ãã¹ãããã¤ã¾ã§ã¼ ã®åç·¨ã§ããããã§ã¯ãäºæ¥å¿ç¨ã«ããã£ã¦Computer Visionæè¡ã«ããDeep Learningï¼ä»¥ä¸ãDLï¼ã¢ãã«ã®è¦ç´ æè¡éçºäºä¾ã«ã¤ãã¦ä»¥ä¸ãããã¯ã§ãç´¹ä»ãã¾ãã 1.
chainer-compilerã®ãã®å¾ Shinichiro Hamaji
AWS Startup ããã° ãéå¬å ±å&è³æå ¬éãML@Loft #10 Edge Deep Learning ããã«ã¡ã¯ãAWS ã½ãªã¥ã¼ã·ã§ã³ã¢ã¼ããã¯ãã®è¾»ã§ãã2020å¹´åã¨ãªã第10åç®ã® ML@Loft ã¯1æã« Deep Learning ãã¬ã¼ã ã¯ã¼ã¯ã¨æ¨è«ããã¼ãã«éå¬ãã¾ããã ML@Loft ã¯æ©æ¢°å¦ç¿ã®ãæ©ã¿ç¸è«ã¤ãã³ãã§ãç®é»ã® AWS Loft Tokyo ã§2019å¹´4æããæ¯æéå¬ãã¦ãã¾ããAWS ãã使ãã®ã客æ§ããµã¼ãã¹ã®ä¸ã«æ©æ¢°å¦ç¿ãåãå ¥ãã¦éçºã»éç¨ãã¦ããéã®ãæ©ã¿ãç¸è«ã§ããå ´ã欲ãããã¨ãããã¨ã§å§ã¾ã£ãã³ãã¥ããã£ã¤ãã³ãã§ããç»å£è ï¼ç¸è«å½¹ï¼ãèªå·±ç´¹ä»ãå ¼ãã10åã»ã©ã® Lighting Talk (LT) ãé çªã«è¡ã£ãå¾ããã¼ãã«ã«åããã¦å ·ä½çãªç¸è«ã»ãã£ã¹ã«ãã·ã§ã³ãè¡ããã¨ããäºé¨æ§æã§éå¬ããã¦ãã¾ããéå»ã®æ§åã¯ç»å£
製é ç¾å ´ã§æ´»èºåºãããç»åÃAIããã©ãããå ´ã§çããã®ãï¼ã¹ãã¼ããã¡ã¯ããªã¼ï¼1/2 ãã¼ã¸ï¼ ã°ã«ã¼ã´ãã¼ãã¯ãã製é Ãç»åAIãããã¼ãã«2020å¹´4æ14æ¥ããªã³ã©ã¤ã³ã»ããã¼ãéå¬ãã°ã«ã¼ã´ãã¼ã ã¢ã«ã¦ã³ãã¨ã°ã¼ã¯ãã£ã 岩éæ¦æ°ã製é æ¥åãã®ã¤ã³ãã©è¨åã»æ©æ¢°ã®ä¿å ¨ã»å®å ¨ç®¡çæ¥åã«ãããç»åAIã®æ´»ç¨äºä¾ãªã©ãç´¹ä»ããã ã°ã«ã¼ã´ãã¼ãã¯ãã製é Ãç»åAIãããã¼ãã«2020å¹´4æ14æ¥ããªã³ã©ã¤ã³ã»ããã¼ãéå¬ãã°ã«ã¼ã´ãã¼ã ã¢ã«ã¦ã³ãã¨ã°ã¼ã¯ãã£ã 岩éæ¦æ°ã製é æ¥åãã®ã¤ã³ãã©è¨åã»æ©æ¢°ã®ä¿å ¨ã»å®å ¨ç®¡çæ¥åã«ãããç»åAIï¼äººå·¥ç¥è½ï¼ã®æ´»ç¨äºä¾ãªã©ãç´¹ä»ããã AIã¨éåã³ã³ãã¥ã¼ãã£ã³ã°æè¡ãæ´»ç¨ããããã¼ã©ã³ãããã¯ã¹ã ã°ã«ã¼ã´ãã¼ãã¯ç¦å²¡å¸ã«æ¬ç¤¾ãç½®ãã½ããã¦ã§ã¢ä¼ç¤¾ã§ã社å¡æ°ã¯ç´50人ãAIãéåã³ã³ãã¥ã¼ã¿ãå©ç¨ããçµã¿åããæé©åã½ãªã¥ã¼ã·ã§ã³ãªã©ãæ
Deleted articles cannot be recovered. Draft of this article would be also deleted. Are you sure you want to delete this article? ã¯ããã« ãã®è¨äºã¯AIã¨ãã¸ã³ã³ãã¹ã( https://signate.jp/competitions/191 )åãã«Xilinxã®CNNæ¨è«ã¨ã³ã¸ã³(DPU)ãUltra96-v2ã§åããã¾ã§ã®è¨é²ã§ãã ï¼ä¸å¿ã³ã³ãã¹ãåãã®ã¬ãã¼ããå ¼ãã¦ãã¾ããï¼ ãã®è¨äºã§ã¯DPUããªãã¹ãç´æ¥è§¦ãæ¹åã§ç°å¢ãä½ãã¾ãã ã«ã¹ã¿ãã¤ãºãä¸è¦ãªå ´åãVitisAIçµç±ã§DPUãã¤ãããæ¹ãLinuxç°å¢ã¨ã¡ã¢ãª16GB以ä¸ã§ã¹ãã¬ã¼ã¸ã®ç©ºããä½è£ãããæ¹ã¯Ultra96V2åãVitis AI(2019.2)ã®çµã¿ç«ã¦æ¹ ãåç §ãã¦ãã ã
ãªã³ã©ã¤ã³éå¬ã¨ãªã£ãCOOL Chips 23 2020å¹´4æ15æ¥ãã17æ¥ã¾ã§ã®3æ¥éãIEEEã主å¬ããã³ã³ãã¥ã¼ã¿ã®å½éå¦ä¼ãCOOL Chips 23ããéå¬ãããã å ã ã¯ãå½å寺ã®æ¥ç«ä¸å¤®ç 究æã§éå¬ã®äºå®ã§ãã£ãããæ°åã³ããã¦ã¤ã«ã¹ã®ææãªã¹ã¯ãæ¸ãããããWebä¸ã«ãããªã³ã©ã¤ã³éå¬ã¨ãªã£ãã使ã£ãã·ã¹ãã ã¯CISCOã®WebEXã§ãæ ¶æ義塾大å¦ã®ã·ã¹ãã ã使ããã¦ããã£ã¦ããã ãªã¢ã«ã®å¦ä¼ã§ãå人ãç¥äººã¨ä¼ã£ã¦ãè¿æ³ãªã©ãä¼è©±ããã¨ãã楽ãã¿ã¯ç¡ããã9æããã®è¬æ¼ã«ã家ã§ãã£ããæé£ãé£ã¹ã¦8æ50åã«ãã½ã³ã³ã®åã«åº§ãã°åå ã§ããã¨ããã®ã¯é常ã«æ¥½ã¡ãã§ããã 主å¬è å´ã®çºè¡¨ã«ããã¨ãCOOL Chips 23ã®åå è ã¯92åã¨ã®ãã¨ã§ãä¾å¹´ã®150åç¨åº¦ã®åå è ã«æ¯ã¹ãã¨2/3ç¨åº¦ã®äººæ°ã§ãã£ãããã¼ãã£ã«éå¬ã§ãããã3ä¸å(IEEEãIEICEããããã¯
LeapMindãCOOL Chips 23ã§èªç¤¾éçºAIæ¨è«ã¢ã¯ã»ã©ã¬ã¼ã¿IPã®æ§è½è¦ç©ããçµæãçºè¡¨æå 端ã®AIãããã®ã¬ã¤ã¢ã¦ãè¨è¨ã»è£½é ãå¾æã¨ããASICã¡ã¼ã«ã¼ã§ããAlchip社ååã®ä¸ãPPAè¦ç©ãããå®æ½ AIï¼Artificial Intelligenceã人工ç¥è½ï¼ã®è¦ç´ æè¡ã§ãããã£ã¼ãã©ã¼ãã³ã°æè¡ãæ´»ç¨ããä¼æ¥ã«åããã½ãªã¥ã¼ã·ã§ã³ãæä¾ããLeapMindæ ªå¼ä¼ç¤¾ï¼æå¨å°ï¼æ±äº¬é½æ¸è°·åºã代表åç· å½¹CEOï¼æ¾ç° ç·ä¸ãèªã¿æ¹ï¼ãªã¼ããã¤ã³ãã以ä¸ï¼LeapMindï¼ã¯ãã³ã³ãã¥ã¼ã¿ã®å½éå¦ä¼ãCOOL Chips 23ãï¼ä¸»å¬: IEEEï¼ã®åºèª¿è¬æ¼ã§ãè¶ ä½æ¶è²»é»åAIæ¨è«ã¢ã¯ã»ã©ã¬ã¼ã¿IPã®éçºæ®µéã§ã®æ§è½è¦ç©ããçµæãçºè¡¨ãã¾ããããªããASICè¨è¨ããã¼ãç¨ããã·ãªã³ã³å®è£ ä½æ¥ã¯ãAlchip Technologies, Limitedï¼æå¨å°ï¼å°æ¹¾çå°å
Deleted articles cannot be recovered. Draft of this article would be also deleted. Are you sure you want to delete this article? ã¯ããã« ã¿ãªãã¾ã¯ããã¾ãã¦ãå½¼æ°ã«ããããªã大å¦ã§åãã¦ããNakaharaã¨ç³ãã¾ããAIã¨ãã¸ã³ã³ãã¹ãï¼ãã¼ãã¦ã§ã¢å®è£ ï¼ãéå¬ããã¦ãã¾ããFPGAã¨ããæ¸ãæãå¯è½LSIã使ã£ãã¬ãã³ã³ãã¼ãã¦ã§ã¢å®è£ ã³ã³ãã¹ãã§è¦æ¦ãã¦ããã¨æãã¾ããç§ã10å¹´ãããFPGAè¨è¨ãã¦ãã¾ãããæªã ã«ãããããã¾ããï¼ç¬ï¼ãã¦ãGPU楽ããã ããã£ã¡ãããã£ãã1 ãã£ã¼ãã©ã¼ãã³ã°ãã£ã¦ããã©ãFPGAã¯åãã¦ãªãã ãããããã¨ããããªããªãã¸ãã¯æ¸ãã¦ãã¾ããã©ãAIããã£ã¦ã¿ããããããã¨ããã½ããã¦ã§ã¢ã¯ç¥ã£ã¦ããã©ããã¼ãã¦ã§ã¢
Google Research ãããTFLite GPU ã®ã¢ã¼ããã¯ãã£è¨è¨ã«ã¤ãã¦ã®è«æã 1907.01989 On-Device Neural Net Inference with Mobile GPUstensorflow/tensorflow/lite/delegates/gpu at master · tensorflow/tensorflow · GitHubãã£ã¨ã¾ã¨ããã¨ã ã¢ãã¤ã« CPU 㧠NN ã®æ¨è«ã¯é£ããè¨ç®éãç±ãããããªã¼ç¹ã«æ¢åã®ãã¸ãã¯ã CPU ã§åãã¦ãã®ã«ãããã«è¨ç®è² è·ã追å ããã¨â¦å®éãFig8 㧠iPhone XS ã® MobileNetV1 ã® CPU æ¨è«ã3 åã¡ããã§ç±ã§ããã©ã¼ãã³ã¹è½ã¡ã¦ãã¢ãã¤ã« GPU ã NN ã®æ¨è«ã«ä½¿ãNPU ãªã©ã®ã¢ã¯ã»ã©ã¬ã¼ã¿ãä¸é¨ã®ãã¤ã¨ã³ãæ©ãããªãããå¤ãã®ããã¤ã¹ãã¿ã¼ã²ããã¨ãããã
ä¸å½ã®IT大æã¢ãªããã°ã«ã¼ããæ¯å¹´éå¬ããããã¯ç³»ã¤ãã³ãã2019æå·ã»é²æ 大ä¼ï¼The Computing Conference 2019ï¼ããéå¹ãã25æ¥ãå社ã®å¼µå»ºéï¼ã¸ã§ãã»ãã£ã³ï¼CTOãç»å£ããå社åã®AIããããå«å 800ï¼HanGuang 800ï¼ããçºè¡¨ããã AIããããéçºããã®ã¯åå°ä½ã«ç¹åããã¢ãªããåä¸ã®ä¼æ¥ãå¹³é å¥åå°ä½ï¼Pingtouge SemiconductorãT-HEADï¼ãã ãå社ã¯æ¨å¹´9æã«è¨ç«ãããããã1å¹´ã®éã«çµã¿è¾¼ã¿ç¨CPUãçé910ï¼XuanTie 910ï¼ãããã³SoCç¨ãã©ãããã©ã¼ã ãç¡å£ï¼Wujianï¼ããçºè¡¨ããã å¼µCTOã«ããã¨ãå«å 800ã¯ResNet-50ï¼ç»åèªèãã¥ã¼ã©ã«ãããã¯ã¼ã¯ï¼ä¸ã®è©¦é¨ã§æé«7ä¸8563 IPSï¼ç§ãããã®ç»ååãè¾¼ã¿æ°ï¼ã®æ¨è«ããã©ã¼ãã³ã¹ãå©ãåºãããããã¯ç¾æ®µéã§æ¥çæé«æ§è½ã¨
Resource-constrained IoT devices, such as sensors and actuators, have become ubiquitous in recent years. This has led to the generation of large quantities of data in real-time, which is an appealing target for AI systems. However, deploying machine learning models on such end-devices is nearly impossible. A typical solution involves offloading data to external computing systems (such as cloud ser
Coral is our new brand for products that provide on-device AI for both prototyping and production projects. It's a platform of hardware components, software tools, and pre-compiled machine learning models, allowing you to create local AI in any form-factor. Coral devices harness the power of Google's Edge TPU machine-learning coprocessor. This is a small ASIC built by Google that's specially-desig
ãªãªã¼ã¹ãé害æ å ±ãªã©ã®ãµã¼ãã¹ã®ãç¥ãã
ææ°ã®äººæ°ã¨ã³ããªã¼ã®é ä¿¡
å¦çãå®è¡ä¸ã§ã
j次ã®ããã¯ãã¼ã¯
kåã®ããã¯ãã¼ã¯
lãã¨ã§èªã
eã³ã¡ã³ãä¸è¦§ãéã
oãã¼ã¸ãéã
{{#tags}}- {{label}}
{{/tags}}