[B! Chainer] rokujyouhitomaã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯

rokujyouhitoma id:rokujyouhitoma

Chainerã«é–¢ã™ã‚‹rokujyouhitomaã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ (23)

${{author_name}}$

{{{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}}

Chainer - Wikipedia
Chainerï¼ˆãƒã‚§ã‚¤ãƒŠãƒ¼ï¼‰ã¯ã€ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã®è¨ˆç®—ãŠã‚ˆã³å¦ç¿’ã‚’è¡Œã†ãŸã‚ã®ã‚ªãƒ¼ãƒ—ãƒ³ã‚½ãƒ¼ã‚¹ã®ã‚½ãƒ•ãƒˆã‚¦ã‚§ã‚¢ãƒ©ã‚¤ãƒ–ãƒ©ãƒªã§ã‚ã‚‹ã€‚ãƒãƒƒã‚¯ãƒ—ãƒãƒ‘ã‚²ãƒ¼ã‚·ãƒ§ãƒ³ï¼ˆèª¤å·®é€†ä¼æ’æ³•ï¼‰ã«å¿…è¦ãªãƒ‡ãƒ¼ã‚¿æ§‹é€ ã‚’ãƒ—ãƒã‚°ãƒ©ãƒ ã®å®Ÿè¡Œæ™‚ã«å‹•çš„ã«ç”Ÿæˆã™ã‚‹ç‰¹å¾´ãŒã‚ã‚Š[4]ã€è¤‡é›‘ãªãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã®æ§‹ç¯‰ã‚’å¿…è¦ã¨ã™ã‚‹ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ï¼ˆæ·±å±¤å¦ç¿’)ã§ç”¨ã„ã‚‰ã‚Œã‚‹[3][1]ã€‚Python 2.xç³»ãŠã‚ˆã³3.xç³»ã‹ã‚‰åˆ©ç”¨ã§ã[è¦å‡ºå…¸]ã€GPUã«ã‚ˆã‚‹æ¼”ç®—ã‚’ã‚µãƒãƒ¼ãƒˆã—ã¦ã„ã‚‹[3][5]ã€‚æ ªå¼ä¼šç¤¾Preferred Networksï¼ˆPFNï¼‰ã‹ã‚‰ãƒªãƒªãƒ¼ã‚¹ã•ã‚Œã¦ã„ã‚‹[5][1]ã€‚2019å¹´12æœˆ5æ—¥ã€é–‹ç™ºå…ƒã®PFNã¯ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯é–‹ç™ºã‚’çµ‚äº†ã—ã¦Chainerã¯ãƒ¡ãƒ³ãƒ†ãƒŠãƒ³ã‚¹ãƒ•ã‚§ãƒ¼ã‚ºã¸ç§»è¡Œã™ã‚‹ã“ã¨ã€è‡ªç¤¾ã¯Chainerã‹ã‚‰FacebookãŒä¸»å°Žã™ã‚‹PyTorchã«é †æ¬¡ç§»è¡Œã™ã‚‹ã“ã¨ã‚’ç™ºè¡¨ã—ãŸ[1]ã€‚ Chainerã¯"define-by-ru
rokujyouhitoma 2018/10/26
Chainer
ãƒªãƒ³ã‚¯
chainer.dataset.concat_examples â€” Chainer 5.0.0 documentation
rokujyouhitoma 2018/10/26
Chainer
ãƒªãƒ³ã‚¯
Chainerã‚’ä½¿ã£ã¦æœ¬æ°—ã§ãƒžãƒ¼ã‚±ãƒƒãƒˆã®äºˆæ¸¬ã‚·ã‚¹ãƒ†ãƒ ã‚’ä½œã£ãŸã‚‰ã“ã†ãªã£ãŸ
200,000 Req/sec ã‚’ã•ã°ãåºƒå‘Šå…¥æœã‚·ã‚¹ãƒ†ãƒ ã‚’æ”¯ãˆã‚‹ãƒ‘ãƒ•ã‚©ãƒ¼ãƒžãƒ³ã‚¹ãƒãƒ¥ãƒ¼ãƒ‹ãƒ³ã‚°è¡“ #jjug_ccc #ccc_g6
rokujyouhitoma 2018/06/11
Chainer
ãƒªãƒ³ã‚¯
æ·±å±¤å¼·åŒ–å¦ç¿’ãƒ©ã‚¤ãƒ–ãƒ©ãƒªChainerRL - Preferred Networks Research & Development
Chainerã‚’ä½¿ã£ãŸæ·±å±¤å¼·åŒ–å¦ç¿’ãƒ©ã‚¤ãƒ–ãƒ©ãƒªChainerRLã‚’å…¬é–‹ã—ã¾ã—ãŸï¼Ž https://github.com/pfnet/chainerrl PFNã‚¨ãƒ³ã‚¸ãƒ‹ã‚¢ã®è—¤ç”°ã§ã™ï¼Žç¤¾å†…ã§Chainerã‚’ä½¿ã£ã¦å®Ÿè£…ã—ã¦ã„ãŸæ·±å±¤å¼·åŒ–å¦ç¿’ã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ ã‚’â€ChainerRLâ€ã¨ã„ã†ãƒ©ã‚¤ãƒ–ãƒ©ãƒªã¨ã—ã¦ã¾ã¨ã‚ã¦å…¬é–‹ã—ã¾ã—ãŸï¼ŽRLã¯Reinforcement Learningï¼ˆå¼·åŒ–å¦ç¿’ï¼‰ã®ç•¥ã§ã™ï¼Žä»¥ä¸‹ã®ã‚ˆã†ãªæœ€è¿‘ã®æ·±å±¤å¼·åŒ–å¦ç¿’ã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ ã‚’å…±é€šã®ã‚¤ãƒ³ã‚¿ãƒ•ã‚§ãƒ¼ã‚¹ã§ä½¿ãˆã‚‹ã‚ˆã†å®Ÿè£…ã—ã¦ã¾ã¨ã‚ã¦ã„ã¾ã™ï¼Ž Deep Q-Network (Mnih et al., 2015) Double DQN (Hasselt et al., 2016) Normalized Advantage Function (Gu et al., 2016) (Persistent) Advantage Learning (Bellemar
rokujyouhitoma 2017/02/22
Chainer
ãƒªãƒ³ã‚¯
åˆå¿ƒè€…ãŒchainerã§ç·šç”»ç€è‰²ã—ã¦ã¿ãŸã€‚ã‚ã‚Šã¨ã§ããŸã€‚
ï¾ƒï¾žï½°ï¾Œï¾Ÿï¾—ï½°ï¾†ï¾ï½¸ï¾žã¯ã‚³ãƒ¢ãƒ‡ã‚£ãƒ†ã‚£åŒ–ã—ã¦ã„ã¦ãƒãƒ³ãƒ€ä»˜ã‘ã®æ–¹ãŒä»˜åŠ ä¾¡å€¤é«˜ã„ã¨ã„ã‚ã‚Œã‚‹ï¾‹ï¾Ÿ-FNã§ä¸»ã«å·¥ä½œæ‹…å½“ã®tai2anã§ã™ã€‚ NHKã§å…¨å›½æ”¾é€ã•ã‚ŒãŸAmazon Picking Challengeã§ã‚¬ãƒ ãƒ†ã¹ã£ãŸã¹ãŸã®ãƒãƒ³ãƒ‰ã‚„ãƒãƒœã‚³ãƒ³æ„Ÿæº€è¼‰ã®æ»‘ã‚Šå°ã¨ã‹ã‚’å·¥ä½œã—ã¦ã¾ã—ãŸã€‚ ã¨ã¯ã„ãˆã€ã‚„ã£ã±ã‚Šã¡ã‚‡ã£ã¨ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ã—ã¦ã¿ãŸã„ã®ã§1,2ã‹æœˆå‰ã‹ã‚‰chainerã‚’å‹‰å¼·ã—å§‹ã‚ã¾ã—ãŸã€‚ ã›ã£ã‹ããªã®ã§ç·šç”»ã®ç€è‰²ã‚’ã—ãŸã„ãªãƒ¼ã¨æ€ã£ã¦è‰²ã€…è©¦ã—ã¦ã¿ã¾ã—ãŸã€‚ ç·šç”»ã®ç€è‰²ã¯æ•™å¸«ã‚ã‚Šå¦ç¿’ãªã®ã§ç·šç”»ã¨ç€è‰²æ¸ˆã¿ã®ç”»åƒã®ãƒ‡ãƒ¼ã‚¿ã‚»ãƒƒãƒˆãŒ(ã§ãã‚Œã°å¤§é‡ã«)å¿…è¦ã§ã™ã€‚ ä»Šå›žã¯OpenCVã§ã‚«ãƒ©ãƒ¼ã®ç”»åƒã‹ã‚‰ç·šç”»ã‚’é©å½“ã«æŠ½å‡ºã—ã¦ã„ã¾ã™ã€‚ æŠ½å‡ºä¾‹ â†’ ã‚«ãƒ©ãƒ¼ã®ç”»åƒã‚’é›†ã‚ã¦ç·šç”»ã‚’ä½œã‚Œã°ãƒ‡ãƒ¼ã‚¿ã‚»ãƒƒãƒˆã®å®Œæˆã§ã™ã€‚ï¼ˆä»Šå›žã¯ï¼–ï¼ä¸‡æžšãã‚‰ã„ä½¿ã£ã¦ã„ã¾ã™ï¼‰ ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã®å½¢ã§ã™ãŒã€U-netã¨ã„ã†æœ€åˆã®æ–¹ã§ã‚³ãƒ³ãƒœãƒªãƒ¥ãƒ¼ã‚·ãƒ§ãƒ³ã™ã‚‹æ™‚ã®å±¤ã®å‡º
rokujyouhitoma 2017/01/28
OpenCV

Chainer
ãƒªãƒ³ã‚¯
PFNã®æ·±å±¤å¦ç¿’ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯ã€ŒChainerã€ãŒå¤§å¹…ãªé«˜é€ŸåŒ–ã¸ã€åˆ†æ•£å‡¦ç†ã«å¯¾å¿œ
Preferred Networksï¼ˆPFNï¼‰ãŒé–‹ç™ºã™ã‚‹ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ï¼ˆæ·±å±¤å¦ç¿’ï¼‰ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯ã€ŒChainerã€ãŒé–“ã‚‚ãªãã€åˆ†æ•£å‡¦ç†ã«å¯¾å¿œã™ã‚‹ã“ã¨ã§å¤§å¹…ã«é«˜é€ŸåŒ–ã™ã‚‹ã€‚PFNã®è¥¿å·å¾¹ç¤¾é•·ãŒ2017å¹´1æœˆ26æ—¥ï¼ˆç±³å›½æ™‚é–“ï¼‰ã«ç±³ã‚µãƒ³ãƒ•ãƒ©ãƒ³ã‚·ã‚¹ã‚³ã§é–‹å‚¬ã•ã‚ŒãŸã€ŒDeep Learning Summit 2017ã€ã§æ˜Žã‚‰ã‹ã«ã—ãŸï¼ˆå†™çœŸ1ï¼‰ã€‚ ç¾åœ¨ã®Chainerã¯ã€ãƒžãƒ«ãƒãƒŽãƒ¼ãƒ‰ï¼ˆè¤‡æ•°ã‚µãƒ¼ãƒãƒ¼ï¼‰ã§ã®å¦ç¿’ã«å¯¾å¿œã«ã—ã¦ãŠã‚‰ãšã€1å°ã®ãƒŽãƒ¼ãƒ‰ä¸Šã§è¤‡æ•°ã®GPUï¼ˆGraphics Processing Unitï¼‰ã‚’ä½¿ç”¨ã™ã‚‹å ´åˆã‚‚ã€ã©ã®GPUã§ä½•ã®å‡¦ç†ã‚’å®Ÿè¡Œã™ã‚‹ã®ã‹ã‚’ãƒ—ãƒã‚°ãƒ©ãƒžãƒ¼ãŒè¨˜è¿°ã™ã‚‹å¿…è¦ãŒã‚ã£ãŸã€‚PFNã¯ãƒžãƒ«ãƒãƒŽãƒ¼ãƒ‰ã§ã®å¦ç¿’ã«å¯¾å¿œã—ãŸåˆ†æ•£ãƒãƒ¼ã‚¸ãƒ§ãƒ³ã®Chainerã‚’é–‹ç™ºä¸ã§ã€ç¤¾å†…ã§ã¯ãƒ†ã‚¹ãƒˆã‚‚å§‹ã‚ã¦ã„ã‚‹ã€‚ã“ã®åˆ†æ•£ãƒãƒ¼ã‚¸ãƒ§ãƒ³ã®Chainerã§ã¯è¨å®šã‚’å¤‰æ›´ã™ã‚‹ã ã‘ã§ã€ãƒžãƒ«ãƒãƒŽãƒ¼ãƒ‰ç’°å¢ƒã‚„ãƒžãƒ«ãƒGPUç’°å¢ƒã«å¯¾å¿œã™ã‚‹
rokujyouhitoma 2017/01/28
PFN

Chainer

PFI
ãƒªãƒ³ã‚¯
æŠ•ç¥¨: ãŒã£ã¤ã‚Šå®Ÿè£…ã€Deep Learning | PyCon JP 2016 in TOKYO
ã¡ã¾ãŸã§ã€ã€ŒDeep Learningã‚¹ã‚²ãƒ¼ã€ã¨ã„ã†å£°ã‚’ã¡ã‚‰ã»ã‚‰èžãã¾ã™ã€‚ã€ŒDeep Learningã‚„ã‚ŠãŸã„ã€ã¨ã„ã†å£°ã‚‚èžãã¾ã™ã€‚ã—ã‹ã—ãã®æ¬¡ã«ã§ã‚‹è¨€è‘‰ãŒã€Œèª°ã‹ã‚„ã£ã¦ã€ã€Œèª°ã‹ã§ãã‚‹äººã„ãªã„ï¼Ÿã€ã¨ã„ã†æ„Ÿã˜ã§ã™ã€‚å°‘ãªãã¨ã‚‚ç§ã®å‘¨ã‚Šã«ã¯ã€‚ ã¾ãŸã€githubã§Deep Learningã®ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯theano ã‚„Chainerã‚’ä½¿ã£ãŸå®Ÿè£…ã‚’è¦‹ã‹ã‘ã¾ã™ãŒã€ç´ æ•µãªå®Ÿè£…ã¯ã‚ã¾ã‚Šè¦‹ã‹ã‘ã¾ã›ã‚“ã€‚ã§ã€æ€ã†ã®ã¯ã€ã¿ã‚“ãªã©ã®ãã‚‰ã„Deep Learningã®Pythonã‚³ãƒ¼ãƒ‰ã‚’ä½œã‚Œã¦ã„ã‚‹ã‚“ã ã‚ã†ï¼Ÿã¨ã„ã†ã“ã¨ã§ã™ã€‚ ãã“ã§ã€theanoã®ãƒ©ãƒƒãƒ‘ãƒ¼ã‚¯ãƒ©ã‚¹ã‚’ã‹ã‚Œã“ã‚Œ10å›žä½ã€ã¤ãã£ã¦ã¯å£Šã—ã‚’ç¹°ã‚Šè¿”ã—ã€Chainerã‚’æœ€è¿‘ä½¿ã„å§‹ã‚ãŸè‘—è€…ãŒã€Chainerã‚’ä½¿ã£ãŸå®Ÿè£…ã«ã¤ã„ã¦ã€ãŒã£ã¤ã‚Šèªžã‚ŠãŸã„ã¨æ€ã„ã¾ã™ã€‚ Deep Learningã®ç ”ç©¶å‹•å‘ã¯ã€ã¨ã‹ã®ã¯ã—ã¾ã›ã‚“ã€‚å®Ÿè£…ã®è©±ã§ã™ã€‚ GPUåŒ–ã™ã‚‹æ–¹æ³•ã¨ãã®é©šãã¹ãåŠ¹æžœã‚‚
rokujyouhitoma 2016/06/15
èˆˆå‘³ã‚ã‚‹ã‘ã©ä¼¼ãŸã®ä¹±ç«‹ã—éŽãŽã§ã€ã©ã‚ŒãŒã„ã„ã®ã‹ã‚ã‹ã‚‰ã‚“ã€‚

Chainer

DeepLearning

MachineLearning

Python

PyCon

PyConJP

PyConJP2016
ãƒªãƒ³ã‚¯
Chainerã®exampleã®mnistã‚’cythonã‚’ä½¿ã£ã¦ã©ã‚Œã ã‘é«˜é€ŸåŒ–ã§ãã‚‹ã‹æ¤œè¨¼ã—ã¦ã¿ãŸ - Qiita
#ãªãœã‚„ã‚‹ã‹ Chainer1.5ã‹ã‚‰é«˜é€ŸåŒ–ã•ã‚Œã¦ã„ã¾ã™ã€‚é«˜é€ŸåŒ–ã®ä¸€å› ã¨ã—ã¦cythonã‚’ä½¿ç”¨ã—ã¦ã„ã‚‹ã“ã¨ãŒæŒ™ã’ã‚‰ã‚Œã¦ã„ãŸã®ã§ã€exampleã‚’cythonã§æ›¸ãæ›ãˆãŸã‚‰ã€ã©ã‚Œã ã‘æ—©ããªã‚‹ã®ã‹å˜ç´”ãªç–‘å•ãŒæ¹§ã„ãŸã®ã§ã€è©¦ã—ã¦ã¿ã¾ã—ãŸã€‚ ##PCã®ã‚¹ãƒšãƒƒã‚¯ OS:OS X Yosem ite CPU: 2.7GHz Intel Core i5 Memory:8GHz DDR3 ##æ¡ä»¶ exampleã®Mnistã‚’ä½¿ç”¨ å¦ç¿’å›žæ•°ï¼š20å›ž ãƒ‡ãƒ¼ã‚¿ã¯äºˆã‚ãƒ€ã‚¦ãƒ³ãƒãƒ¼ãƒ‰ã•ã‚ŒãŸçŠ¶æ…‹ ##å¯è¦–åŒ– ãƒ—ãƒãƒ•ã‚¡ã‚¤ãƒ©ã«ã‚ˆã‚‹å¯è¦–åŒ– pycallgraphã‚’ä½¿ç”¨ http://pycallgraph.slowchop.com graphivzã‚’ã‚¤ãƒ³ã‚¹ãƒˆãƒ¼ãƒ« http://www.graphviz.org/Download_macos.php X11ã‚’ã‚¤ãƒ³ã‚¹ãƒˆãƒ¼ãƒ«(Yosem iteã®å ´åˆ) http://www.xqua
rokujyouhitoma 2015/12/22
Chainer

Cython
ãƒªãƒ³ã‚¯
Installation â€” Chainer 7.8.1 documentation
InstallationÂ¶ Recommended EnvironmentsÂ¶ We recommend the following Linux distributions. Ubuntu 14.04 / 16.04 LTS (64-bit) CentOS 7 (64-bit) Note We are automatically testing Chainer on all the recommended environments above. We cannot guarantee that Chainer works on other environments including Windows and macOS (especially with CUDA support), even if Chainer may seem to be running correctly. Requ
rokujyouhitoma 2015/12/16
Chainer
ãƒªãƒ³ã‚¯
Chainer â€“ A flexible framework of neural networks â€” Chainer 7.8.1 documentation
Chainer â€“ A flexible framework of neural networksÂ¶ Chainer is a powerful, flexible and intuitive deep learning framework. Chainer supports CUDA computation. It only requires a few lines of code to leverage a GPU. It also runs on multiple GPUs with little effort. Chainer supports various network architectures including feed-forward nets, convnets, recurrent nets and recursive nets. It also supports
rokujyouhitoma 2015/12/16
Chainer
ãƒªãƒ³ã‚¯
Chainer 1.5ã®ã‚¤ãƒ³ã‚¹ãƒˆãƒ¼ãƒ«ãŒã†ã¾ãã„ã‹ãªã„äººã¸ã®éžå…¬å¼ãªTips - Qiita
Chainer 1.6ä»¥é™ã§ã€ã“ã®è¨˜äº‹ã®å•é¡Œã®å¤§éƒ¨åˆ†ã¯è§£æ¶ˆã—ã¾ã—ãŸ Chainer 1.5ã§ã€å¤–éƒ¨ã®ä¾å˜çŠ¶æ³ãŒå¤§ããå¤‰ã‚ã‚Šã¾ã—ãŸã€‚ãã®çµæžœã€å¹¾ã¤ã‹ã®è¦å› ã§ã‚¤ãƒ³ã‚¹ãƒˆãƒ¼ãƒ«ãŒé›£ã—ããªã‚Šã¾ã—ãŸã€‚ã“ã®è¨˜äº‹ã§ã¯ã€ãã®å¯¾ç–ã‚’æ›¸ãã¾ã™ã€‚ ä¸»ãªåŽŸå› Cythonã‚’ä½¿ã†ã‚ˆã†ã«ãªã‚Šã¾ã—ãŸã€‚ã‚½ãƒ¼ã‚¹ã®ä¸€éƒ¨ã¯.pyxãƒ•ã‚¡ã‚¤ãƒ«ã«ãªã‚Šã€.cppã‚’ç”Ÿæˆã—ã¦.soã«ãƒ“ãƒ«ãƒ‰ã—ã¦ã„ã¾ã™ï¼ˆ1.5.1ã§cythonizeæ¸ˆã¿ã®.cppãƒ•ã‚¡ã‚¤ãƒ«ã‚’åŒæ¢±ã—ã¦ã„ã¾ã™ï¼‰ CUDAã‚„cuDNNãªã©ã®å…±æœ‰ãƒ©ã‚¤ãƒ–ãƒ©ãƒªï¼ˆ.soãƒ•ã‚¡ã‚¤ãƒ«ï¼‰ã‚‚Cythonã‹ã‚‰ä½¿ã†ã‚ˆã†ã«ãªã‚Šã¾ã—ãŸã€‚ä»Šã¾ã§ã¯ã€ctypesã‚’ä½¿ã£ã¦ã„ã¾ã—ãŸã€‚ h5pyãªã©ã€å…±æœ‰ãƒ©ã‚¤ãƒ–ãƒ©ãƒªã‚’å¿…è¦ã¨ã™ã‚‹ãƒ‘ãƒƒã‚±ãƒ¼ã‚¸ã‚’åˆ©ç”¨ã™ã‚‹ã‚ˆã†ã«ãªã‚Šã¾ã—ãŸã€‚ é †ã«ã€ä½•ãŒå•é¡Œã«ãªã‚‹ã®ã‹èª¬æ˜Žã—ã¾ã™ã€‚ Cythonã®åˆ©ç”¨ Cythonã¯.pyxã§è¨˜è¿°ã—ãŸãƒ•ã‚¡ã‚¤ãƒ«ã‚’ã€.cppã«å¤‰æ›ï¼ˆã‚³ãƒ³ãƒ‘ã‚¤ãƒ«ï¼‰ã—ã¦ã‹ã‚‰ã€.soãƒ•ã‚¡ã‚¤ãƒ«ã‚’ä½œ
rokujyouhitoma 2015/12/14
Chainer
ãƒªãƒ³ã‚¯
Chainerã«ã‚ˆã‚‹å¤šå±¤ãƒ‘ãƒ¼ã‚»ãƒ—ãƒˆãƒãƒ³ã®å®Ÿè£… - äººå·¥çŸ¥èƒ½ã«é–¢ã™ã‚‹æ–å‰µéŒ²
ã“ã‚Œã¾ã§Deep Learningã®ã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ ã‚’Theanoã§å®Ÿè£…ã—ã¦ããŸï¼ˆ2015/4/29ï¼‰ã‘ã‚Œã©ã€ã“ã“ã‚‰ã§å··ã§å¤§äººæ°—ã®ãƒ©ã‚¤ãƒ–ãƒ©ãƒªChainerã«ã‚‚æ‰‹ã‚’å‡ºã—ã¦ã¿ãŸã€‚Theanoã®å‹‰å¼·ã‚’å§‹ã‚ãŸã‚ã¨ã™ãã«ChainerãŒå…¬é–‹ã•ã‚ŒãŸï¼ˆ2015/6/9ï¼‰ãŒãƒ¦ãƒ¼ã‚¶ã‚„æƒ…å ±ãŒå¢—ãˆã‚‹ã¾ã§å¾…ã£ã¦ã„ãŸæ„Ÿã˜ï¼ˆç¬‘ï¼‰æœ€è¿‘ã¯ã‚³ãƒ¼ãƒ‰ã‚„å®Ÿé¨“çµæžœãªã©ã‚’å…¬é–‹ã—ã¦ãã‚Œã‚‹äººãŒå¢—ãˆã¦ããŸã®ã§éžå¸¸ã«å‚è€ƒã«ãªã£ã¦ã„ã‚‹ã€‚ç›®ã«ã¤ã„ãŸã‚‚ã®ã¯ã¦ã¶ã«ç™»éŒ²ã—ã¦ã„ã‚‹ã®ã§ã€èˆˆå‘³ã‚’æŒã£ãŸæ‰‹æ³•ã¯ãŒã—ãŒã—å‹‰å¼·ã—ã¦è¿½è©¦ã—ã¦ã„ããŸã„ã€‚ Chainerã®ãƒãƒ¼ã‚¸ãƒ§ãƒ³ã¯1.3.2ã‚’ãƒ™ãƒ¼ã‚¹ã«ã—ã¦ã„ã‚‹ã€‚1.3ã‹ã‚‰PyCUDA/scikit-cudaã‚’ç‹¬è‡ªãƒ©ã‚¤ãƒ–ãƒ©ãƒªã®CuPyã«ç½®ãæ›ãˆãŸã¨ã®ã“ã¨ã§ã€ä»¥å‰ã®ã‚³ãƒ¼ãƒ‰ã¯å°‘ã—ä¿®æ£ã—ãªã„ã¨å‹•ã‹ãªã„ã‚ˆã†ã ã€‚ãã®åˆ†ã€1.3ã‹ã‚‰ã¯ã‚¤ãƒ³ã‚¹ãƒˆãƒ¼ãƒ«ãŒã‚·ãƒ³ãƒ—ãƒ«ã«ãªã£ã¦ã„ã¦ã¨ã¦ã‚‚ã†ã‚Œã—ã„ã€‚1.1ã®ã“ã‚ã¯ã€Chainerã¨ç›´æŽ¥é–¢ä¿‚ãªã„PyCUD
rokujyouhitoma 2015/10/06
Chainer
ãƒªãƒ³ã‚¯
Chainerã®ä½¿ã„æ–¹ã¨è‡ªç„¶è¨€èªžå‡¦ç†ã¸ã®å¿œç”¨
2. â¾ƒè‡ªâ¼°å·±ç´¹ä»‹ lï¬â€¯ å¾—å±… Â èª ä¹Ÿ Â (Seiya Â Tokui) Â Â @beam2d Â (Twitter, Â GitHub) lï¬â€¯ PFI Â (2012-â€â€‘â€’2014) Â -â€â€‘â€’> Â PFN Â (2014-â€â€‘â€’) lï¬â€¯ å°‚â¾¨é–€ï¼šæ©Ÿæ¢°å¦ç¿’ â€“â€¯ â¾ƒè‡ªç„¶â¾”è¨€èªžå‡¦ç†ï§¤ï¼ˆå“è©žã‚¿ã‚°ä»˜ã‘ï¼‰ã€è¿‘å‚æŽ¢ç´¢ï¥ªã®ãŸã‚ã®ãƒãƒƒã‚·ãƒ³ã‚°ï¼ˆä¿®â¼ å£«ï¼‰ â€“â€¯ Deep Â Learningï¼ˆ2012å¹´ï¦Žç§‹ã”ã‚ã€œï½žï¼‰ã€ç”»åƒãƒ»æ˜ åƒèªè˜ï§¼ lï¬â€¯ 4 Â â½‰æœˆã« Â Chainer Â ã‚’ä½œã‚Šå§‹ã‚ã¦ä»Šã¯ã“ã‚ŒãŒãƒ¡ã‚¤ãƒ³ 2 3. Chainer ã‚µã‚¤ãƒˆ: Â http://chainer.org ãƒ‰ã‚ãƒ¥ãƒ¡ãƒ³ãƒˆ: Â http://docs.chainer.org GitHub: Â https://github.com/pfnet/chainer lï¬â€¯ Deep Â Learning Â ã®ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯ lï¬â€¯ Python Â
rokujyouhitoma 2015/09/04
Chainer

NLP
ãƒªãƒ³ã‚¯
http://yans.anlp.jp/modules/journal/journal_detail.php?block_id=28&news_id=76&op=comment
rokujyouhitoma 2015/08/18
Chainer

NLP
ãƒªãƒ³ã‚¯
Deep Learningã‚’ç”¨ã„ãŸå°†æ£‹ãƒ—ãƒã‚°ãƒ©ãƒ Gunyanzaã‚’å…¬é–‹ã—ã¾ã—ãŸ
ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°(æ·±å±¤å¦ç¿’)ã¨ã„ã†ã®ãŒæµè¡Œã£ã¦ã„ã‚‹ãã†ã§ã™ã€‚ã™ã¹ã¦ã®äººé¡žã¯ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ã«ã‚ˆã£ã¦å®Ÿç¾ã•ã‚ŒãŸAIã«éš·å±žã™ã‚‹æœªæ¥ãªã‚“ã§ã™ã£ã¦ã‚ˆ!!! ã“ã‚ãƒ¼ã„ã€‚ ãã‚“ãªãƒãƒ©è‰²ã®æŠ€è¡“ã€ã„ã£ã¡ã‚‡ã‹ã¿ã—ã¦ãŠããŸã„ã§ã™ã‚ˆãã€‚ ã•ã¦ã€ã‚ªãƒ•ã‚£ã‚¹ã§ç¤¾é•·ã¨ãƒ€ãƒ™ã£ã¦ã„ãŸã¨ã“ã‚ã€ã€Œå°†æ£‹ãƒ—ãƒã‚°ãƒ©ãƒ é¢ç™½ãã†ã ã‚ˆãã€ã¨ã„ã†è©±ã«ãªã‚Šã¾ã—ãŸã€‚ãŠäº’å°†æ£‹ãƒ—ãƒã‚°ãƒ©ãƒ ã‚’ä½œã£ã¦é—˜ã‚ã›ã‚ˆã†ãœã€ã„ã‡ãƒ¼ã€ã¨ç››ã‚Šä¸ŠãŒã‚Šã€å‹¢ã„ã§ã‚³ãƒ³ãƒ”ãƒ¥ãƒ¼ã‚¿å°†æ£‹é¸æ‰‹æ¨©ã«ç”³ã—è¾¼ã¿ã¾ã§ã—ã¦ã—ã¾ã„ã¾ã—ãŸã€‚ ãã‚“ãªçµŒç·¯ã§ã€ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ã‚’ãƒŸãƒªã—ã‚‰(=1ãƒŸãƒªã‚‚çŸ¥ã‚‰ãªã„)ãªåƒ•ãŒã€è©¦ã—ã«ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ã‚’ä½¿ã£ã¦å°†æ£‹ã®AIã‚’æ›¸ã„ã¦ã¿ãŸã‚‰ã„ã„ã‚„ã‚“ã€ã¨æ€ã„ã¤ã„ãŸã®ã§ã—ãŸã€‚å°†æ£‹ã‚‚ã€ãƒãƒ å°†æ£‹ã§ãƒãƒ 8æžšè½ã¡ã§è² ã‘ã‚‹ãƒ¬ãƒ™ãƒ«ãã‚‰ã„ã€‚ãƒ€ãƒ¡ã˜ã‚ƒã‚“ã€‚ ãƒŸãƒªã—ã‚‰ãªã®ã§ã€ã€Œãƒã‚§ã‚¹ã§ä½•ã‹ã‚„ã£ã¦ã‚‹ãƒ¤ãƒ„ãŒã„ã‚‹ã ã‚ã†ã€ã¨ã‚¢ã‚¿ãƒªã‚’ã¤ã‘ã¦ã‚°ã‚°ã£ãŸçµæžœã€Erik Bernhardssonã•ã‚“ã«ã‚ˆã‚‹ d
rokujyouhitoma 2015/08/06
Gunyanza

DeepLearning

å°†æ£‹

gunyarakun

Chainer
ãƒªãƒ³ã‚¯
ã€æ©Ÿæ¢°å¦ç¿’ã€‘ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚° ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯Chainerã‚’è©¦ã—ãªãŒã‚‰è§£èª¬ã—ã¦ã¿ã‚‹ã€‚ - Qiita
ä»Šè©±é¡Œã®Deep Learning(æ·±å±¤å¦ç¿’)ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯ã€Chainerã«æ‰‹æ›¸ãæ–‡å—ã®åˆ¤åˆ¥ã‚’è¡Œã†ã‚µãƒ³ãƒ—ãƒ«ã‚³ãƒ¼ãƒ‰ãŒã‚ã‚Šã¾ã™ã€‚ã“ã¡ã‚‰ã‚’ä½¿ã£ã¦å†…å®¹ã‚’å°‘ã—è§£èª¬ã™ã‚‹è¨˜äº‹ã‚’æ›¸ã„ã¦ã¿ãŸã„ã¨æ€ã„ã¾ã™ã€‚ (æœ¬è¨˜äº‹ã®ã‚³ãƒ¼ãƒ‰ã®å…¨æ–‡ã‚’GitHubã«ã‚¢ãƒƒãƒ—ã—ã¾ã—ãŸã€‚[PCæŽ¨å¥¨]) ã¨ã«ã‹ãã€ã‚¤ãƒ³ã‚¹ãƒˆãƒ¼ãƒ«ãŒã™ã”ãç°¡å˜ã‹ã¤ã€PythonãŒæ›¸ã‘ã‚Œã°ã™ãã«ä½¿ã†ã“ã¨ãŒã§ãã¦ãŠã™ã™ã‚ã§ã™ï¼ Pythonã«é–‰ã˜ã¦ã‚³ãƒ¼ãƒ‰ãŒæ›¸ã‘ã‚‹ã®ã‚‚ã™ã”ãã„ã„ã§ã™ã‚ˆãã€‚ ã“ã‚“ãªæ„Ÿã˜ã®ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ãƒ¢ãƒ‡ãƒ«ã‚’è©¦ã—ã¦ã¿ã‚‹ã€ã¨ã„ã†è¨˜äº‹ã§ã™ã€‚ ä¸»è¦ãªæƒ…å ±ã¯ã“ã¡ã‚‰ã«ã‚ã‚Šã¾ã™ã€‚ Chainerã®ãƒ¡ã‚¤ãƒ³ã‚µã‚¤ãƒˆ Chainerã®GitHubãƒªãƒã‚¸ãƒˆãƒª Chainerã®ãƒãƒ¥ãƒ¼ãƒˆãƒªã‚¢ãƒ«ã¨ãƒªãƒ•ã‚¡ãƒ¬ãƒ³ã‚¹ #1. ã‚¤ãƒ³ã‚¹ãƒˆãƒ¼ãƒ«# ã¾ãšã¯ä½•ã¯ã¨ã‚‚ã‚ã‚Œã‚¤ãƒ³ã‚¹ãƒˆãƒ¼ãƒ«ã§ã™ã€‚Chainerã®GitHubã«è¨˜è¼‰ã®"Requirements" ( https://github.
rokujyouhitoma 2015/06/23
Chainer
ãƒªãƒ³ã‚¯
DeepLearningãƒ©ã‚¤ãƒ–ãƒ©ãƒªã®ChainerãŒã™ã”ã„ã€ã‚‰ã—ã„ - cvl-robot's diary
Preferred Infrastructure(ä»¥ä¸‹PFI)ã‹ã‚‰ã‚¹ãƒ”ãƒ³ã‚ªãƒ•ã—ãŸä¼šç¤¾ã€Preferred Networksã®ãƒªãƒªãƒ¼ã‚¹ã—ãŸDeepLearningãƒ©ã‚¤ãƒ–ãƒ©ãƒªã®ChainerãŒã™ã”ã„ã€ã¨è©±é¡Œã«ãªã£ã¦ã„ã¾ã™ã€‚*1 è§£èª¬ Deep Learning ã®ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯ Chainer ã‚’å…¬é–‹ã—ã¾ã—ãŸ | Preferred Research å…¬å¼ Chainer: A flexible framework of neural networks GitHub pfnet/chainer Â· GitHub ãƒ‰ã‚ãƒ¥ãƒ¡ãƒ³ãƒˆ Chainer â€“ A flexible framework of neural networks â€” Chainer 1.1.0 documentation ãŠãã‚‰ãåˆéœ²å‡º ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°æœ€è¿‘ã®ç™ºå±•ã¨ãƒ“ã‚¸ãƒã‚¹å¿œç”¨ã¸ã®èª²é¡Œ å…¬å¼ãƒ„ã‚¤ãƒƒã‚¿ãƒ¼ chainer (@Chai
rokujyouhitoma 2015/06/21
Chainer

DeepLearning

MachineLearning
ãƒªãƒ³ã‚¯
ã€æ©Ÿæ¢°å¦ç¿’ã€‘ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚° ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯Chainerã‚’è©¦ã—ãªãŒã‚‰è§£èª¬ã—ã¦ã¿ã‚‹ã€‚ - Qiita
ä»Šè©±é¡Œã®Deep Learning(æ·±å±¤å¦ç¿’)ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯ã€Chainerã«æ‰‹æ›¸ãæ–‡å—ã®åˆ¤åˆ¥ã‚’è¡Œã†ã‚µãƒ³ãƒ—ãƒ«ã‚³ãƒ¼ãƒ‰ãŒã‚ã‚Šã¾ã™ã€‚ã“ã¡ã‚‰ã‚’ä½¿ã£ã¦å†…å®¹ã‚’å°‘ã—è§£èª¬ã™ã‚‹è¨˜äº‹ã‚’æ›¸ã„ã¦ã¿ãŸã„ã¨æ€ã„ã¾ã™ã€‚ (æœ¬è¨˜äº‹ã®ã‚³ãƒ¼ãƒ‰ã®å…¨æ–‡ã‚’GitHubã«ã‚¢ãƒƒãƒ—ã—ã¾ã—ãŸã€‚[PCæŽ¨å¥¨]) ã¨ã«ã‹ãã€ã‚¤ãƒ³ã‚¹ãƒˆãƒ¼ãƒ«ãŒã™ã”ãç°¡å˜ã‹ã¤ã€PythonãŒæ›¸ã‘ã‚Œã°ã™ãã«ä½¿ã†ã“ã¨ãŒã§ãã¦ãŠã™ã™ã‚ã§ã™ï¼ Pythonã«é–‰ã˜ã¦ã‚³ãƒ¼ãƒ‰ãŒæ›¸ã‘ã‚‹ã®ã‚‚ã™ã”ãã„ã„ã§ã™ã‚ˆãã€‚ ã“ã‚“ãªæ„Ÿã˜ã®ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ãƒ¢ãƒ‡ãƒ«ã‚’è©¦ã—ã¦ã¿ã‚‹ã€ã¨ã„ã†è¨˜äº‹ã§ã™ã€‚ ä¸»è¦ãªæƒ…å ±ã¯ã“ã¡ã‚‰ã«ã‚ã‚Šã¾ã™ã€‚ Chainerã®ãƒ¡ã‚¤ãƒ³ã‚µã‚¤ãƒˆ Chainerã®GitHubãƒªãƒã‚¸ãƒˆãƒª Chainerã®ãƒãƒ¥ãƒ¼ãƒˆãƒªã‚¢ãƒ«ã¨ãƒªãƒ•ã‚¡ãƒ¬ãƒ³ã‚¹ #1. ã‚¤ãƒ³ã‚¹ãƒˆãƒ¼ãƒ«# ã¾ãšã¯ä½•ã¯ã¨ã‚‚ã‚ã‚Œã‚¤ãƒ³ã‚¹ãƒˆãƒ¼ãƒ«ã§ã™ã€‚Chainerã®GitHubã«è¨˜è¼‰ã®"Requirements" ( https://github.
rokujyouhitoma 2015/06/19
Chainer

MachineLearning

DeepLearning

Python
ãƒªãƒ³ã‚¯
Chainer: A flexible framework for neural networks
A Powerful, Flexible, and Intuitive Framework for Neural Networks Get Started Learn More
rokujyouhitoma 2015/06/09
Chainer

DeepLearning

MachineLearning
ãƒªãƒ³ã‚¯
Chainer â€“ A flexible framework of neural networks â€” Chainer 7.8.1 documentation
Chainer â€“ A flexible framework of neural networksÂ¶ Chainer is a powerful, flexible and intuitive deep learning framework. Chainer supports CUDA computation. It only requires a few lines of code to leverage a GPU. It also runs on multiple GPUs with little effort. Chainer supports various network architectures including feed-forward nets, convnets, recurrent nets and recursive nets. It also supports
rokujyouhitoma 2015/06/09
chainer

PFI

DeepLearning

MachineLearning
ãƒªãƒ³ã‚¯
1 2 æ¬¡ã®ãƒšãƒ¼ã‚¸