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

tune id:tune

deeplearningã«é–¢ã™ã‚‹tuneã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ (4)

${{author_name}}$

{{{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}}

Linux Foundationã€æ©Ÿæ¢°å¦ç¿’ã‚’ã‚¢ãƒ—ãƒªã‚±ãƒ¼ã‚·ãƒ§ãƒ³ã«ç°¡å˜ã«çµ„ã¿è¾¼ã‚ã‚‹ã‚ªãƒ¼ãƒ—ãƒ³ã‚½ãƒ¼ã‚¹ã€ŒAcumos AIã€ç™ºè¡¨ã€‚æ–°å›£ä½“ã€ŒLF Deep Learning Foundationã€ãŒé–‹ç™º
Linux Foundationã€æ©Ÿæ¢°å¦ç¿’ã‚’ã‚¢ãƒ—ãƒªã‚±ãƒ¼ã‚·ãƒ§ãƒ³ã«ç°¡å˜ã«çµ„ã¿è¾¼ã‚ã‚‹ã‚ªãƒ¼ãƒ—ãƒ³ã‚½ãƒ¼ã‚¹ã€ŒAcumos AIã€ç™ºè¡¨ã€‚æ–°å›£ä½“ã€ŒLF Deep Learning Foundationã€ãŒé–‹ç™º Acumos AI ãƒ—ãƒã‚¸ã‚§ã‚¯ãƒˆã¯ãŠã‚‚ã«ã€ŒAcumosãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯ã€ã¨ã€ŒAcumosãƒžãƒ¼ã‚±ãƒƒãƒˆãƒ—ãƒ¬ã‚¤ã‚¹ã€ã‹ã‚‰æ§‹æˆã•ã‚Œã€ã“ã‚Œã‚‰ã«ã‚ˆã£ã¦AIã‚„æ©Ÿæ¢°å¦ç¿’ã®é–‹ç™ºã€åˆ©ç”¨ã€å…±æœ‰ã¨ã„ã†3ã¤ã®ãƒ¡ãƒªãƒƒãƒˆã‚’ã‚‚ãŸã‚‰ã™ã¨ã—ã¦ã„ã¾ã™ã€‚ ä»¥ä¸‹ã¯Acumos AI ãƒ—ãƒã‚¸ã‚§ã‚¯ãƒˆã®ãƒ–ãƒã‚°ã«æŠ•ç¨¿ã•ã‚ŒãŸè¨˜äº‹ã€ŒWhat is Acumos? - Acumosã€ã‹ã‚‰ã€‚ Acumosã¯ã€AIãƒ¢ãƒ‡ãƒ«ã®é–‹ç™ºè€…ã«å¯¾ã—ã¦ã¯å®¹æ˜“ãªé–‹ç™ºã¨ãƒ‘ãƒƒã‚±ãƒ¼ã‚¸ãƒ³ã‚°ã‚’å®Ÿç¾ã™ã‚‹ã‚½ãƒ•ãƒˆã‚¦ã‚§ã‚¢ã ã¨èª¬æ˜Žã•ã‚Œã¦ã„ã¾ã™ã€‚ For creators of AI models â€“ data scientists and others â€“ Acumos empowers them to
tune 2018/04/09
æœ€åˆã‹ã‚‰ãƒžãƒ¼ã‚±ãƒƒãƒˆãƒ—ãƒ¬ã‚¤ã‚¹ã‚’ç”¨æ„ã—ã¡ã‚ƒã†ã‚ˆã†ãªãƒ—ãƒãƒ€ã‚¯ãƒˆã¯ãã‚Œã»ã©ã¯ã‚„ã‚‰ãªã„æ°—ãŒã™ã‚‹

DeepLearning

OSS

AI
ãƒªãƒ³ã‚¯
LSTMã§ãƒã‚¤ãƒŠãƒªãƒ‡ãƒ¼ã‚¿ã‚’èªã‚€~ã‚ã‚‹ã„ã¯ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã«ã‚ˆã‚‹JPEGã®å†è©•ä¾¡~ - Qiita
ã“ã®è¨˜äº‹ã§ã¯ä»¥ä¸‹ã®å†…å®¹ã‚’ã¯ãªã—ã¾ã™ ãƒã‚¤ãƒŠãƒªãƒ‡ãƒ¼ã‚¿ã‚’ãã®ã¾ã¾ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã«çªã£è¾¼ã‚€ã“ã¨ãŒã§ãã‚‹ ã‚°ãƒªãƒƒãƒJPEGç”»åƒã‚‚èªè˜ã§ãã‚‹ CNN+LSTMã®æ§‹é€ ãŒã‚ˆããƒã‚¤ãƒŠãƒªãƒ‡ãƒ¼ã‚¿ã‚’å¦ç¿’ã™ã‚‹ JPEGã¯ãƒãƒã‚¹ãƒˆãªç”»åƒçš„ç‰¹å¾´ã‚’æ‰ãˆã‚„ã™ã„ãƒã‚¤ãƒŠãƒªãƒ•ã‚©ãƒ¼ãƒžãƒƒãƒˆã§ã‚ã‚‹ ãƒã‚¤ãƒŠãƒªãƒ‡ãƒ¼ã‚¿èªè˜ã¨ç”»åƒèªè˜ã®é•ã„ å¦ä¼šç™ºè¡¨ã®ãŸã‚ãƒãƒ³ã‚³ã‚¯ã«æ¥ã¦ã„ã¾ã™ã€‚ @Hi-king ã§ã™ã€‚ã“ã®è¨˜äº‹ã¯ ãƒ‰ãƒ¯ãƒ³ã‚´ Advent Calendar 2017 åˆæ—¥ã®è¨˜äº‹ã§ã™ã€‚ æ—©é€Ÿã§ã™ãŒã€ä»Šæ—¥ã®ãƒ†ãƒ¼ãƒžã®èƒŒæ™¯ã®ãƒã‚¨ãƒ ã‚’èªã¿ã¾ã™ã€‚ã‚³ãƒ³ãƒ”ãƒ¥ãƒ¼ã‚¿ãƒ“ã‚¸ãƒ§ãƒ³ã¯æ©Ÿæ¢°ã®ç›®ã‚’ä½œã‚‹å¦å•ã ã¨è¨€ã‚ã‚Œã¦ã„ã¦ï¼Œç‰¹ã«è¿‘å¹´ã®ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ã®æŠ€è¡“ã«ã‚ˆã‚Šï¼Œ"ç‰¹ã«å‰å‡¦ç†ã¨ã‹ã—ãªãã¦ã‚‚ã€ç”Ÿã®ç”»åƒã‚’ç›´æŽ¥ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆã«å…¥åŠ›ã™ã‚Œã°ç”»åƒèªè˜ã§ãã‚‹"ã¨ã„ã†èƒ½åŠ›ã‚’ç²å¾—ã—ãŸã¨è¨€ã‚ã‚Œã¦ã„ã¾ã™ã€‚ ã—ã‹ã—ã€ã¡ã‚‡ã£ã¨å¾…ã£ã¦ä¸‹ã•ã„ã€‚æˆ‘ã€…ã‚¨ãƒ³ã‚¸ãƒ‹ã‚¢ã«ã¨ã£ã¦ã€ç”Ÿã®ç”»åƒãƒ‡ãƒ¼ã‚¿ã£ã¦æœ¬å½“ã«ç”»
tune 2017/12/02
ä½Žå‘¨æ³¢æˆåˆ†ã‚’ä¸å¿ƒã«ä½¿ã£ã¦ã‚‹ã‚“ã ã‚ã†ãªã€‚H.264ã‚‚ãƒ‡ã‚³ãƒ¼ãƒ‰ã›ãšã«èªè˜å‡¦ç†ã«å›žã›ã‚‹ã‚ˆã†ã«ãªã‚‹ã¨è‰²ã€…è‰¯ã•ãã†

jpeg

æ©Ÿæ¢°å¦ç¿’

DeepLearning
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‚¯
AIã¯å‹˜é•ã„ã™ã‚‹â€•ãƒ†ã‚¹ãƒ©ã¨ãƒˆãƒ¨ã‚¿ã«ã¿ã‚‹è‡ªå‹•é‹è»¢ã‚«ãƒ¼æˆ¦ç•¥ã®é•ã„
ã“ã®ãƒ©ãƒ³ãƒ€ãƒ ãƒŽã‚¤ã‚ºã®ã‚ˆã†ãªç”»åƒã‚’ã€æ‰‹æ›¸ãã®æ•°å—ï¼ˆ0ï½ž9ï¼‰ã‚’99.8ï¼…ã®ç²¾åº¦ã§èªè˜ã§ãã‚‹äººå·¥çŸ¥èƒ½ï¼ˆã“ã®å ´åˆã¯ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆï¼‰ã«å…¥åŠ›ã™ã‚‹ã¨ã€ãªã‚“ã¨äººå·¥çŸ¥èƒ½ã¯ã»ã¼100ï¼…ã®è‡ªä¿¡åº¦ã§ã€Œ1ã€ã¨åˆ¤åˆ¥ã—ã¦ã—ã¾ã†ã€‚ ã“ã®ç”»åƒã¯ã€ç‰¹å®šã®ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆã‚’ã€Œã ã¾ã™ã€ç›®çš„ã§ã€AI æŠ€è¡“é–‹ç™ºã®ã‚«ãƒ©ãƒ•ãƒ«ãƒ»ãƒœãƒ¼ãƒ‰ãŒä½œæˆã—ãŸã‚‚ã®ã ã€‚å¦ç¿’æ¸ˆã¿ã®ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆã«ç”»åƒã‚’å…¥åŠ›ã—ã€ãã®å‡ºåŠ›çµæžœãŒã€Œ1ã€ã«è¿‘ã¥ãã‚ˆã†ã€ç”»åƒã«ç¹°ã‚Šè¿”ã—ãƒ•ã‚£ãƒ¼ãƒ‰ãƒãƒƒã‚¯ã‚’ä¸ŽãˆãŸã‚‚ã®ã§ã‚ã‚‹ã€‚ åŒç¤¾CEOã®æ¸¡è¾ºç¥æ¨¹æ°ã¯ã€Œãªãœãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãŒã“ã®ãƒ©ãƒ³ãƒ€ãƒ ãªç”»åƒã‚’ã€Ž1ã€ã¨èªè˜ã™ã‚‹ã®ã‹ã€ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆã®æŒ™å‹•ã‹ã‚‰è§£æžã™ã‚‹ã“ã¨ã¯å›°é›£ã€ã¨èªžã‚‹ã€‚ä¸€èˆ¬çš„ãªãƒ—ãƒã‚°ãƒ©ãƒ ã‚³ãƒ¼ãƒ‰ã¨ç•°ãªã‚Šã€ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆã¯å¤§é‡ã®ãƒ‘ãƒ©ãƒ¡ãƒ¼ã‚¿ã‹ã‚‰æ§‹æˆã•ã‚Œã€äººé–“ã«ã¨ã£ã¦å¯èªæ€§ã®ä½Žã„ãƒ–ãƒ©ãƒƒã‚¯ãƒœãƒƒã‚¯ã‚¹ã ã‹ã‚‰ã ã€‚ äººå·¥çŸ¥èƒ½ã®åˆ¤æ–èª¤ã‚ŠãŒäº‹æ•…ã‚’ç”Ÿã‚€ äººå·¥çŸ¥èƒ½ã¯ã€å¾€ã€…ã«ã—ã¦ã ã¾ã•ã‚Œã€å‹˜é•ã„ã—ã€é–“é•ã£ãŸåˆ¤
tune 2016/07/14
è‡ªå‹•è»Š

DeepLearning

ãƒˆãƒ¨ã‚¿
ãƒªãƒ³ã‚¯
1