[B! deep learning] tuneã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯

tune id:tune

deep learningã«é–¢ã™ã‚‹tuneã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ (8)

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

${{author_name}}$
{{author_name}}{{created}}
{{ #comment }}{{ comment }}{{ /comment }}
- {{ label }}

${{author_name}}$

{{{comment_expanded}}}

{{label}}

{{#is_bookmark}}ãƒªã‚¹ãƒˆ{{/is_bookmark}}{{^is_bookmark}}ãƒªãƒ³ã‚¯{{/is_bookmark}}

Amazon Kinesis Video Streams Ã— Deep Learning - ABEJA Tech Blog
ã‚¨ãƒ³ã‚¸ãƒ‹ã‚¢ã®æ²³å´Žã§ã™ã€‚ AWS re:Invent2017ã§ã€æ–°ã‚µãƒ¼ãƒ“ã‚¹ã®Kinesis Video Streamã®ã‚¤ãƒ³ãƒˆãƒãƒ€ã‚¯ã‚·ãƒ§ãƒ³ã‚»ãƒƒã‚·ãƒ§ãƒ³ã§ãƒ‡ãƒ¢ã¨ãƒ¦ãƒ¼ã‚¹ã‚±ãƒ¼ã‚¹ã®ç™ºè¡¨ã‚’ã•ã›ã¦ã„ãŸã ãã¾ã—ãŸã€‚ www.youtube.com ã“ã®ã€Kinesis Video Streamsã®ç´¹ä»‹ã¨re:Inventã§ç™ºè¡¨ã—ãŸãƒ‡ãƒ¢ã®æ§‹æˆã‚’JAWS-UG AIã§ç™ºè¡¨ã•ã›ã¦ã„ãŸã ãã¾ã—ãŸã®ã§ã€ã”è¦§ã«ãªã£ã¦ãã ã•ã„ã€‚ speakerdeck.com å®£ä¼ æ¥å¹´2æœˆã«ABEJA SIX 2018ã¨ã„ã†ã‚¤ãƒ™ãƒ³ãƒˆã‚’é–‹å‚¬ã—ã¾ã™ã€‚æ˜ åƒé–¢ä¿‚ã®äº‹ä¾‹ã®ç™ºè¡¨ã‚‚ã‚ã‚Šã¾ã™ã®ã§æ˜¯éžã”å‚åŠ ãã ã•ã„ã€‚ six.abejainc.com è£½é€ æ¥ã‚„ã‚¤ãƒ³ãƒ•ãƒ©ã€æµé€šå°å£²æ¥ãªã©ã§ã€äººå·¥çŸ¥èƒ½ã‚’ã©ã®ã‚ˆã†ã«æ´»ç”¨ã—ã¦ã„ã‚‹ã‹ã€å®Ÿäº‹ä¾‹ã‚’å¤šæ•°ã”ç´¹ä»‹ã™ã‚‹ï¼‘DAYã‚«ãƒ³ãƒ•ã‚¡ãƒ¬ãƒ³ã‚¹ã§ã™ã€‚ æœ¬ã‚«ãƒ³ãƒ•ã‚¡ãƒ¬ãƒ³ã‚¹ã§ã¯ã€äººå·¥çŸ¥èƒ½æ´»ç”¨ã‚’æŽ¨é€²ã—ã¦äº‹æ¥åŠ¹çŽ‡åŒ–ã€ç”Ÿç”£æ€§å‘ä¸Šã«æˆåŠŸã—ãŸå„æ¥ç•Œ
tune 2017/12/16
ãƒ‡ã‚³ãƒ¼ãƒ‰è‡ªå‰ã¨ã‹ãƒ‡ãƒãƒƒã‚°ã¤ã‚‰ã„ã¨ã‹ã‚„ã£ã±ã‚Šã‚ã‚‹ã‚ˆãã€‚ç¾å®Ÿã¯åŽ³ã—ã„ã€‚

amazon

Deep Learning

AWS

Amazon Kinesis
ãƒªãƒ³ã‚¯
ã‚¿ã‚«ãƒã‚·æ˜¥ã® GAN ç¥ã‚Šï¼ã€œ ä¸€æ—¥ä¸€GAN(à¹‘â€¢Ì€ã…‚â€¢Ì)Ùˆâœ§ ã€œ - ABEJA Tech Blog
ABEJAã§ãƒªã‚µãƒ¼ãƒãƒ£ãƒ¼ã‚’ã—ã¦ã„ã¾ã™é«˜æ©‹ã§ã™ã€‚ æ˜¨ä»Š deep learning ç•Œéšˆã§ã¯ Generative Adversarial Net(GAN) ãŒæµè¡Œã£ã¦ã„ã¦ã€ä¸–ã¯ã¾ã•ã«ã‚¬ãƒ³ã‚¬ãƒ³è¡Œã“ã†ãœæ™‚ä»£ã§ã™ãã€‚ GAN ã‚’ç”¨ã„ã‚‹ã¨ç¶ºéº—ãªçµµãŒä½œæˆã§ããŸã‚ŠäºŒã¤ã®çµµã®ä¸é–“ã®ã‚ˆã†ãªçµµã‚’ç”Ÿæˆã§ããŸã‚Šã—ã¾ã™ã€‚ä¾‹ãˆã°ã“ã®è«–æ–‡ã®ã‚ˆã†ãªæ„Ÿã˜ã§ã™ã€‚ã“ã®ã‚ˆã†ã« GAN ã¯æœ‰ç”¨ãªãƒ¢ãƒ‡ãƒ«ã§ã‚ã‚‹ä¸€æ–¹ã€æœ€è¿‘ã® GAN ã§ã¯æ€¥ã«ã‚ˆãã‚ã‹ã‚‰ãªã„å¼ãŒå‡ºã¦ããŸã‚Šã™ã‚‹ã®ã§ã€å‹‰å¼·ã‚‚å…¼ãã¦ã€Œä¸€æ—¥ä¸€GANã€ã‚’ã‚„ã£ã¦ã¿ã¾ã—ãŸã€‚ä»Šå›žèªã‚“ã è«–æ–‡ã®ãƒªã‚¹ãƒˆã¯ä»¥ä¸‹ã§ã™ã€‚ EBGAN (https://arxiv.org/abs/1609.03126 ) WGAN (https://arxiv.org/abs/1701.07875) LSGAN (https://arxiv.org/abs/1611.04076) f-GAN (https://ar
tune 2017/05/31
deep learning

æ©Ÿæ¢°å¦ç¿’

GAN
ãƒªãƒ³ã‚¯
ç•³ã¿è¾¼ã¿ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã®ä»•çµ„ã¿ | POSTD
(ç·¨æ³¨ï¼š2016/11/17ã€è¨˜äº‹ã‚’ä¿®æ£ã„ãŸã—ã¾ã—ãŸã€‚) ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ã®åˆ†é‡Žã§ãƒ†ã‚¯ãƒŽãƒã‚¸ã®é€²åŒ–ãŒç¶šã„ã¦ã„ã‚‹ã¨ã„ã†ã“ã¨ãŒè©±é¡Œã«ãªã‚‹å ´åˆã€åä¸å…«ä¹ç•³ã¿è¾¼ã¿ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ãŒé–¢ä¿‚ã—ã¦ã„ã¾ã™ã€‚ç•³ã¿è¾¼ã¿ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã¯CNNï¼ˆConvolutional Neural Networkï¼‰ã¾ãŸã¯ConvNetã¨ã‚‚å‘¼ã°ã‚Œã€ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã®åˆ†é‡Žã®ä¸»åŠ›ã¨ãªã£ã¦ã„ã¾ã™ã€‚CNNã¯ç”»åƒã‚’è¤‡æ•°ã®ã‚«ãƒ†ã‚´ãƒªã«åˆ†é¡žã™ã‚‹ã‚ˆã†å¦ç¿’ã—ã¦ãŠã‚Šã€ãã®åˆ†é¡žèƒ½åŠ›ã¯äººé–“ã‚’ä¸Šå›žã‚‹ã“ã¨ã‚‚ã‚ã‚Šã¾ã™ã€‚å¤§è¨€å£®èªžã®ã†ãŸã„æ–‡å¥ã‚’å®Ÿç¾ã—ã¦ã„ã‚‹æ–¹æ³•ãŒæœ¬å½“ã«ã‚ã‚‹ã¨ã™ã‚Œã°ã€ãã‚Œã¯CNNã§ã—ã‚‡ã†ã€‚ CNNã®éžå¸¸ã«å¤§ããªé•·æ‰€ã¨ã—ã¦ã€ç†è§£ã—ã‚„ã™ã„ã“ã¨ãŒæŒ™ã’ã‚‰ã‚Œã¾ã™ã€‚å°‘ãªãã¨ã‚‚å¹¾ã¤ã‹ã®åŸºæœ¬çš„ãªéƒ¨åˆ†ã«ãƒ–ãƒ¬ãƒ¼ã‚¯ãƒ€ã‚¦ãƒ³ã—ã¦å¦ã¹ã°ã€ãã‚Œã‚’å®Ÿæ„Ÿã§ãã‚‹ã§ã—ã‚‡ã†ã€‚ã¨ã„ã†ã‚ã‘ã§ã€ã“ã‚Œã‹ã‚‰ä¸€é€šã‚Šèª¬æ˜Žã—ã¾ã™ã€‚ã¾ãŸã€ç”»åƒå‡¦ç†ã«ã¤ã„ã¦ã“ã®è¨˜äº‹ã‚ˆã‚Šã‚‚è©³ç´°ã«èª¬æ˜Ž
tune 2016/11/20
åˆ†ã‹ã‚Šã‚„ã™ã„ã€ã‘ã©é›£ã—ã„ã€‚

deeplearning

æ©Ÿæ¢°å¦ç¿’

deep learning
ãƒªãƒ³ã‚¯
Machine Intelligence at Google Scale: Vision/Speech API, TensorFlow and Cloud Machine Learning
The biggest challenge of Deep Learning techno logy is the scalability. As long as using single GPU server, you have to wait for hours or days to get the â€¦
tune 2016/04/14
google

TensorFlow

Deep Learning
ãƒªãƒ³ã‚¯
ç”»é¢¨ã‚’å¤‰æ›ã™ã‚‹ã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ - Preferred Networks Research & Development
Deep Neural Networkã‚’ä½¿ã£ã¦ç”»åƒã‚’å¥½ããªç”»é¢¨ã«å¤‰æ›ã§ãã‚‹ãƒ—ãƒã‚°ãƒ©ãƒ ã‚’Chainerã§å®Ÿè£…ã—ã€å…¬é–‹ã—ã¾ã—ãŸã€‚ https://github.com/mattya/chainer-gogh ã“ã‚“ã«ã¡ã¯ã€PFNãƒªã‚µãƒ¼ãƒãƒ£ãƒ¼ã®æ¾å…ƒã§ã™ã€‚ãƒ–ãƒã‚°ã®1è¡Œç›®ã¯botã«æŒã£ã¦è¡Œã‹ã‚Œã‚„ã™ã„ã®ã§ã€3è¡Œç›®ã§æŒ¨æ‹¶ã—ã¦ã¿ã¾ã—ãŸã€‚ ä»Šå›žå®Ÿè£…ã—ãŸã®ã¯â€A Neural Algorithm of Artistic Styleâ€(å…ƒè«–æ–‡)ã¨ã„ã†ã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ ã§ã™ã€‚ç”Ÿæˆã•ã‚Œã‚‹ç”»åƒã®ç¾Žã—ã•ã¨ã€ç”»åƒèªè˜ã®ã‚¿ã‚¹ã‚¯ã§äºˆã‚è¨“ç·´ã—ãŸãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆã‚’ãã®ã¾ã¾æµç”¨ã§ãã‚‹ã¨ã„ã†ãŠæ‰‹è»½ã•ã‹ã‚‰ã€ä¸–ç•Œä¸ã§è©±é¡Œã«ãªã£ã¦ã„ã¾ã™ã€‚ã“ã®ã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ ã®ä»•çµ„ã¿ãªã©ã‚’èª¬æ˜Žã—ãŸã„ã¨æ€ã„ã¾ã™ã€‚ æ¦‚è¦ 2æžšã®ç”»åƒã‚’å…¥åŠ›ã—ã¾ã™ã€‚ç‰‡æ–¹ã‚’ã€Œã‚³ãƒ³ãƒ†ãƒ³ãƒ„ç”»åƒã€ã€ã‚‚ã†ç‰‡æ–¹ã‚’ã€Œã‚¹ã‚¿ã‚¤ãƒ«ç”»åƒã€ã¨ã—ã¾ã—ã‚‡ã†ã€‚ ã“ã®ãƒ—ãƒã‚°ãƒ©ãƒ ã¯ã€ã‚³ãƒ³ãƒ†ãƒ³ãƒ„ç”»åƒã«æ›¸ã‹ã‚ŒãŸç‰©ä½“ã®é…ç½®ã‚’ãã®ã¾
tune 2015/09/14
Deep Learning

ç”»åƒå‡¦ç†

æ©Ÿæ¢°å¦ç¿’
ãƒªãƒ³ã‚¯
Deep Learning ã®ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯ Chainer ã‚’å…¬é–‹ã—ã¾ã—ãŸ - Preferred Networks Research & Development
ã“ã‚“ã«ã¡ã¯ã€å¾—å±…ã§ã™ã€‚æœ€è¿‘ã¯æ¯Žæ™©ã‚¤ã‚«ã«ãªã£ã¦æˆ¦å ´ã‚’é§†ã‘ã¾ã‚ã£ã¦ã„ã¾ã™ã€‚ æœ¬æ—¥ã€Deep Learning ã®æ–°ã—ã„ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯ã§ã‚ã‚‹ Chainer ã‚’å…¬é–‹ã—ã¾ã—ãŸã€‚ Chainer å…¬å¼ã‚µã‚¤ãƒˆ GitHub â€“ pfnet/chainer Chainer Documentation Chainer ã¯ã€ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆã‚’èª¤å·®é€†ä¼æ’æ³•ã§å¦ç¿’ã™ã‚‹ãŸã‚ã®ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯ã§ã™ã€‚ä»¥ä¸‹ã®ã‚ˆã†ãªç‰¹å¾´ã‚’æŒã£ã¦ã„ã¾ã™ã€‚ Python ã®ãƒ©ã‚¤ãƒ–ãƒ©ãƒªã¨ã—ã¦æä¾›ï¼ˆè¦ Python 2.7+ï¼‰ ã‚ã‚‰ã‚†ã‚‹ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆã®æ§‹é€ ã«æŸ”è»Ÿã«å¯¾å¿œ å‹•çš„ãªè¨ˆç®—ã‚°ãƒ©ãƒ•æ§‹ç¯‰ã«ã‚ˆã‚‹ç›´æ„Ÿçš„ãªã‚³ãƒ¼ãƒ‰ GPU ã‚’ã‚µãƒãƒ¼ãƒˆã—ã€è¤‡æ•° GPU ã‚’ã¤ã‹ã£ãŸå¦ç¿’ã‚‚ç›´æ„Ÿçš„ã«è¨˜è¿°å¯èƒ½ ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆã‚’ã©ã®ã‚ˆã†ã«æ›¸ã‘ã‚‹ã‹ æ¬¡ã®ã‚³ãƒ¼ãƒ‰ã¯å¤šå±¤ãƒ‘ãƒ¼ã‚»ãƒ—ãƒˆãƒãƒ³ã®å‹¾é…ã‚’è¨ˆç®—ã™ã‚‹ä¾‹ã§ã™ã€‚ from chainer import FunctionSet, Vari
tune 2015/06/12
Deep Learning

ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯
ãƒªãƒ³ã‚¯
https://jp.techcrunch.com/2015/01/17/20150116facebook-open-sources-some-of-its-deep-learning-tools/
tune 2015/01/17
Facebook

deep learning

æ©Ÿæ¢°å¦ç¿’
ãƒªãƒ³ã‚¯
Deep Learningã¨ç”»åƒèªè˜ï¿½ã€€ï¿½ï½žæ´å²ãƒ»ç†è«–ãƒ»å®Ÿè·µï½ž
1. Nakayama Lab. Machine Perception Group The University of Tokyo æ±äº¬å¤§å¦ å¤§å¦é™¢æƒ…å ±ç†å·¥å¦ç³»ç ”ç©¶ç§‘ å‰µé€ æƒ…å ±å¦å°‚æ”» ä¸å±±ç ”ç©¶å®¤ ä¸å±± è‹±æ¨¹ 2. Nakayama Lab. Machine Perception Group The University of Tokyo ï½ ï¼‘ï¼ŽDeep learningï¼ˆæ·±å±¤å¦ç¿’ï¼‰ã¨ã¯ï¼Ÿ ï½ ï¼’ï¼Žä¸€èˆ¬ç”»åƒèªè˜ï¼šDeep learning ä»¥å‰ã¨ä»¥å¾Œã§ä½•ãŒå¤‰ã‚ã£ãŸã‹ â—¦ Bag-of-visual-words (VLAD, Fisher Vector) â—¦ Convolutional neural network (ConvNets) ï½ ï¼“ï¼ŽDeep learningã®æ•°ç† â—¦ ãªãœå„ªã‚ŒãŸæ€§èƒ½ãŒå®Ÿç¾ã§ãã‚‹ã®ã‹ï¼Ÿ â—¦ ãƒ–ãƒ¬ãƒ¼ã‚¯ã‚¹ãƒ«ãƒ¼ã‚’ç”Ÿã‚“ã å„è¦ç´ æŠ€è¡“ â—¦ ä¸å±±ç ”ç©¶å®¤ã§ã®ç ”ç©¶ç´¹ä»‹ ï½ ï¼”ï¼Žå®Ÿ
tune 2014/11/02
é›£ã—ã„ã€å¾Œã§èªã¿ç›´ãã†

ç”»åƒèªè˜

Deep Learning

æ©Ÿæ¢°å¦ç¿’
ãƒªãƒ³ã‚¯
1