[B! chainer][Deep learning] p_tanã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯

p_tan id:p_tan

chainerã¨Deep learningã«é–¢ã™ã‚‹p_tanã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ (10)

${{author_name}}$

{{{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}}

Chainerã§Deep Image Priorã‚’ã‚„ã£ã¦ã¿ãŸ - Qiita
Chainer Advent Calendar 2017ã®17æ—¥ç›®ã§ã™ã€‚ ã¯ã˜ã‚ã« æž ã«ç©ºããŒã§ããŸã®ã§ã€æœ€è¿‘ã®å®Ÿè£…ã—ã¦è¦‹ãŸè«–æ–‡ã®ç°¡å˜ãªè§£èª¬ã¨çµæžœã‚’ç´¹ä»‹ã—ã¾ã™ã€‚ä»Šæ—¥ã‚‚ä»Šæ—¥ã¨ã¦ã€äººã®è¤Œã§ç›¸æ’²ã‚’å–ã£ã¦ã„ã“ã†ã¨æ€ã„ã¾ã™ã€‚ ä»Šæ—¥ç´¹ä»‹ã™ã‚‹è«–æ–‡ã¯ã€ŒDeep Image Priorã€ã¨ã„ã†ã‚‚ã®ã§ã™ã€‚ç”»åƒå‡¦ç†ã®åŸºæœ¬ã‚¿ã‚¹ã‚¯ã¨ã—ã¦ã€ãƒŽã‚¤ã‚ºé™¤åŽ»ãƒ»è¶…è§£åƒãƒ»ã‚¤ãƒ³ãƒšã‚¤ãƒ³ãƒ†ã‚£ãƒ³ã‚°ãªã©ã€è‰²ã€…ãªæ‰‹æ³•ãŒç ”ç©¶ã•ã‚Œã¦ã„ã¾ã™ãŒã€Deep Image Priorã¯ã€ã“ã‚Œã‚‰ã®ã‚¿ã‚¹ã‚¯ã«æ±Žç”¨çš„ã«ä½¿ãˆã‚‹ä¸ã€…ä¾¿åˆ©ãªãƒ†ã‚¯ãƒ‹ãƒƒã‚¯ã§ã™ã€‚ ä»¥ä¸‹ã®å›³ãŒè‘—è€…ã®è«–æ–‡ã§æ‰±ã‚ã‚Œã¦ã„ã‚‹ã‚¿ã‚¹ã‚¯ç¾¤ã§ã™ã€‚è‰²ã€…ä½¿ãˆã¦ã™ã”ã„ã§ã™ãã€‚ ã¨ã„ã†ã“ã¨ã§ã€æœ¬æ—¥ã¯ã“ã‚Œã‚’å®Ÿè£…ã—ã¦ã„ãã¾ã—ã‚‡ã†ã€‚ å‰æçŸ¥è˜ ä¸Šè¨˜ã«æŒ™ã’ãŸç”»åƒå‡¦ç†ç¾¤ã¯ã€åŸºæœ¬çš„ã«ã¯ä»¥ä¸‹ã®å¼ã‚’æœ€å°åŒ–ã™ã‚‹ã“ã¨ã§è¡Œã‚ã‚Œã¾ã™ã€‚ $x_0$ã¯å…ƒã®ç”»åƒã§ã‚ã‚Šã€ä¾‹ãˆã°ãƒŽã‚¤ã‚ºãªã©ã‚’å«ã‚€ç”»åƒã€$x$ã¯ç”Ÿæˆç”»åƒã¨ã—ã¾ã™ã€‚$E$ã¯å…ƒã®ç”»åƒã¨å‡¦ç†å¾Œ
p_tan 2018/09/10
chainer

ç”»åƒ

Deep Image Prior

Deep learning

CNN
ãƒªãƒ³ã‚¯
Deconvolutionã®ä»£ã‚ã‚Šã«Pixel Shufflerã‚’ä½¿ã†
æ¦‚è¦ Chainerã§Pixel Shufflerã‚’å®Ÿè£…ã—ãŸ ã¯ã˜ã‚ã« æ˜¨å¹´ã«Twitterç¤¾ãŒç™ºè¡¨ã—ãŸè¶…è§£åƒã®è«–æ–‡ã‚’èªã‚“ã§ã„ãŸã¨ã“ã‚ã€ç”»åƒã®æ‹¡å¤§ã«ã‚ˆãç”¨ã„ã‚‰ã‚Œã‚‹Deconvolutionã‚’ä½¿ã‚ãšã€Pixel Shufflerã¨å‘¼ã°ã‚Œã‚‹ä»•çµ„ã¿ã‚’åˆ©ç”¨ã—ã¦è¶…è§£åƒã‚’è¡Œã£ã¦ãŠã‚Šã€æ°—ã«ãªã£ãŸã®ã§èª¿ã¹ã¾ã—ãŸã€‚ Pixel Shufflerã¯Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Networkã§ææ¡ˆã•ã‚ŒãŸã‚‚ã®ã§ï¼ˆæ£å¼ã«ã¯Sub-Pixel Convolutionï¼‰ã€å…¥åŠ›ç‰¹å¾´ãƒžãƒƒãƒ—ã®å„ãƒ”ã‚¯ã‚»ãƒ«ã‚’ä¸¦ã¹æ›¿ãˆã¦é«˜è§£åƒåº¦ãªç‰¹å¾´ãƒžãƒƒãƒ—ã‚’å‡ºåŠ›ã—ã¾ã™ã€‚ ä»Šå›žã¯DCGANã®Generatorã‚’Deconvolutionã¨Pixel Shufflerã®ä¸¡æ–¹ã§æ§‹æˆã—
p_tan 2017/03/20
è¶…è§£åƒ

chainer

dcgan

deep learning

deeplearning

CNN

Pixel Shuffler

è«–æ–‡
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
p_tan 2017/02/21
å¼·åŒ–å¦ç¿’

deep learning

æ©Ÿæ¢°å¦ç¿’

deeplearning

ãƒ©ã‚¤ãƒ–ãƒ©ãƒª

chainer
ãƒªãƒ³ã‚¯
chainerMNã«ã‚ˆã‚‹åˆ†æ•£å¦ç¿’
p_tan 2017/02/02
ChainerMNã®åˆ†æ•£å‡¦ç†æˆ¦ç•¥ã¯åŒæœŸåž‹ã®ãƒ‡ãƒ¼ã‚¿ä¸¦åˆ—ã€MPIã‚’ä½¿ç”¨ã€‚ä»–ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯ã¨ã®å¯¾æ¯”ãŒé¢ç™½ã‹ã£ãŸã€‚ãƒ™ãƒ³ãƒãƒžãƒ¼ã‚¯ã®å·®ã¯å‡ä¸€æ€§èƒ½ã®GPUç¾¤ã¨é«˜é€ŸãªInfiniBandã ã‹ã‚‰ã“ãã®ã‚ˆã†ãªæ°—ã‚‚ã™ã‚‹ã€‚æ€§èƒ½ã®ã°ã‚‰ã¤ã„ãŸãƒŽãƒ¼ãƒ‰ã ã¨ã©ã†ã‹ï¼Ÿ

deep learning

deeplearning

chainer

chainermn

åˆ†æ•£å‡¦ç†
ãƒªãƒ³ã‚¯
DNC (Differentiable Neural Computers) ã®æ¦‚è¦ + Chainer ã«ã‚ˆã‚‹å®Ÿè£… - Qiita
ã“ã‚“ã«ã¡ã¯ï¼Ž@yos1up ã¨ç”³ã—ã¾ã™ï¼Žå¥½ããªé£Ÿã¹ç‰©ã¯ï¼Œç·‘è‰²ã®ã‚ãƒŽã‚³ã§ã™ï¼Ž 2016/10/12ã« DeepMind ãŒ Nature ã«æŠ•ç¨¿ã—ãŸè«–æ–‡ï¼ŒHybrid computing using a neural network with dynamic external memory ã«ã¦ææ¡ˆã•ã‚Œã¦ã„ã‚‹ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ãƒ¢ãƒ‡ãƒ« DNC (Differentiable Neural Computers) ã‚’ï¼Œå¤§æ€¥ãŽã§ Chainer ã§å®Ÿè£…ã—ã¦ã¿ã¾ã—ãŸï¼Ž DNC ã«ã¤ã„ã¦ DNC ã¯ï¼Œå…ˆè¿°ã®è«–æ–‡ã§ææ¡ˆã•ã‚ŒãŸæ–°ã—ã„ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã§ï¼Œãã®æƒ…å ±å‡¦ç†èƒ½åŠ›ã®é«˜ã•ãŒæœŸå¾…ã•ã‚Œã¦ã„ã¾ã™ï¼Žè«–æ–‡ä¸ã§ã¯ï¼Œã‚°ãƒ©ãƒ•ä¸Šã®æœ€çŸçµŒè·¯ã‚¿ã‚¹ã‚¯ã‚„ã¡ã‚‡ã£ã¨ã—ãŸãƒ‘ã‚ºãƒ«ã®ã‚¿ã‚¹ã‚¯ãªã©ï¼Œå¾“æ¥ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã§ã¯å¦ç¿’ä¸å¯èƒ½ã¨æ€ã‚ã‚Œã¦ã„ãŸã‚¿ã‚¹ã‚¯ãŒ DNC ã§å¦ç¿’ã§ãã¦ãŠã‚Šï¼Œãã®æƒ…å ±å‡¦ç†èƒ½åŠ›ã®é«˜ã•ãŒã†ã‹ãŒã„çŸ¥ã‚Œã¾ã™ï¼Ž DN
p_tan 2016/10/26
DeepLearning

deep learning

chainer

Python

æ©Ÿæ¢°å¦ç¿’

deeplearning

Differentiable Neural Computers

DNC
ãƒªãƒ³ã‚¯
ç”Ÿæˆãƒ¢ãƒ‡ãƒ«ã® Deep Learning
PFIã‚»ãƒŸãƒŠãƒ¼(2016/02/25)ã§ç™ºè¡¨ã—ãŸã‚¹ãƒ©ã‚¤ãƒ‰ã§ã™ã€‚ä¼æ‰¿ã‚µãƒ³ãƒ—ãƒªãƒ³ã‚°å¯èƒ½ãªç”Ÿæˆãƒ¢ãƒ‡ãƒ«ã«é–¢ã™ã‚‹Deep Learningã®ç´¹ä»‹ã§ã™ï¼ˆã‚ãƒ¼ãƒ¯ãƒ¼ãƒ‰ï¼šWake-Sleep, å¤‰åˆ† AutoEncoder, Generative Adversarial Nets, Likelihood Ratioï¼‰Read less
p_tan 2016/10/04
ç”Ÿæˆãƒ¢ãƒ‡ãƒ«

deep learning

deeplearning

chainer

ã‚¹ãƒ©ã‚¤ãƒ‰

PFN
ãƒªãƒ³ã‚¯
ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ã®æœ‰åãƒ©ã‚¤ãƒ–ãƒ©ãƒª5ç¨®ã‚’æœ€çŸè·é›¢ã§è©¦ã™åŠæ—¥ã‚³ãƒ¼ã‚¹ï¼ˆTensorFlow, Chainer, Caffe, DeepDream, ç”»é¢¨å¤‰æ›ï¼‰ - ãã®å¾Œã®ãã®å¾Œ
ã€Œã„ã¤ã‹å‹‰å¼·ã—ã‚ˆã†ã€ã¨äººå·¥çŸ¥èƒ½ï¼æ©Ÿæ¢°å¦ç¿’ï¼ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ï¼ˆDeep Learningï¼‰ã¨ã„ã£ãŸãƒˆãƒ”ãƒƒã‚¯ã®è¨˜äº‹ã®è¦‹ã¤ã‘ã¦ã¯ã‚¢ãƒ¼ã‚«ã‚¤ãƒ–ã—ã¦ããŸã‚‚ã®ã®ã€çµå±€2015å¹´ã¯ä½•ä¸€ã¤ã‚„ã‚‰ãšã«çµ‚ã‚ã£ã¦ã—ã¾ã£ãŸã®ã§ã€ã¨ã«ã‹ãä¸€æ©ã§ã‚‚è¶³ã‚’è¸ã¿å‡ºã™ã¹ãã€æœ¬è³ªçš„ãªç†è§£ç‰ã¯ã•ã¦ãŠãã€ã¨ã«ã‹ãè©¦ã—ã¦ã¿ã‚‹ã¨ã„ã†ã“ã¨ã‚’ã‚„ã£ã¦ã¿ã¾ã—ãŸã€‚ è©¦ã—ãŸã®ã¯ã€TensorFlowã€Chainerã€Caffe ã¨ã„ã£ãŸæ©Ÿæ¢°å¦ç¿’ãŠã‚ˆã³ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ã®ä»£è¡¨çš„ãªãƒ©ã‚¤ãƒ–ãƒ©ãƒªï¼ãƒ•ãƒ¬ãƒ¼ãƒ ãƒ¯ãƒ¼ã‚¯3ç¨®ã¨ã€2015å¹´ã«è©±é¡Œã«ãªã£ãŸãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ã‚’åˆ©ç”¨ã—ãŸã‚¢ãƒ—ãƒªã‚±ãƒ¼ã‚·ãƒ§ãƒ³2ç¨®ï¼ˆDeepDreamã€chainer-goghï¼‰ã€‚ ï¼ˆDeepDreamã§è©¦ã—ãŸçµæžœç”»åƒï¼‰ ã‚¿ã‚¤ãƒˆãƒ«ã«åŠæ—¥ã¨æ›¸ãã¾ã—ãŸãŒã€ãŸã¨ãˆã°TensorFlowã¯ç’°å¢ƒæ§‹ç¯‰ã ã‘ãªã‚‰10åˆ†ã‚‚ã‚ã‚Œã°çµ‚ã‚ã‚‹ã§ã—ã‚‡ã†ã—ã€Chainerãªã‚“ã¦ã‚³ãƒžãƒ³ãƒ‰ä¸€ç™ºãªã®ã§5ç§’ãã‚‰ã„ã§ã™ã€‚Caffeã¯åƒ•ã¯ãƒ
p_tan 2016/01/06
ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°

deep learning

chainer

deeplearning

ã¾ã¨ã‚
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
p_tan 2015/09/11
ã¨ã¦ã‚‚è‰¯ãåˆ†ã‹ã‚‹è§£èª¬ã ã£ãŸã€‚å‡„ã„ã€‚

chainer

æ©Ÿæ¢°å¦ç¿’

Deep Learning

ç”»åƒå‡¦ç†

algorithm

ã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ
ãƒªãƒ³ã‚¯
DQNã®ç”Ÿã„ç«‹ã¡ã€€ï¼‹ã€€Deep Q-Networkã‚’Chainerã§æ›¸ã„ãŸ - Qiita
#ã¯ã˜ã‚ã« ãã‚‚ãã‚‚DQNãŒä½œã‚ŠãŸã‹ã£ãŸã‚ã‘ã˜ã‚ƒãªãã¦ã€ä»–ã®ç›®çš„ã§Chainerã‚’ä½¿ã„ãŸã‹ã£ãŸã®ã§ãã®ç·´ç¿’ã«DQNã‚’æ›¸ã„ãŸã‚“ã§ã™ãŒã€ã›ã£ã‹ãã ã—å…¬é–‹ã—ã‚ˆã†ã¨æ€ã„ã¾ã—ãŸ å…¬é–‹ã—ã¾ã—ãŸ ã€‚ã¾ãŸã©ã†ã›å…¬é–‹ã™ã‚‹ãªã‚‰ã€ã“ã®æ©Ÿä¼šã«ã“ã‚Œï¼ˆQå¦ç¿’+é–¢æ•°è¿‘ä¼¼ï¼‰é–¢é€£ã§æŒã£ã¦ã„ã‚‹çŸ¥è˜ã‚’ã¤ã„ã§ã«æ•´ç†ã—ã‚ˆã†ã¨æ€ã£ã¦ã¾ã¨ã‚ã¾ã—ãŸã€‚ ãƒ‹ãƒ¥ãƒ¼ã‚¹è¨˜äº‹ã¨ã‹Natureã¨ã‹NIPSã®è«–æ–‡ã ã‘èªã‚€ã¨ã€DQNä½œã£ãŸDeepmind/Googleã™ã’ãƒ¼ï¼ï¼ï¼ã£ã¦æ„Ÿè¦šã«ãªã‚Šãã†ã§ã™ãŒã€å¼·åŒ–å¦ç¿’çš„ãªæ´å²çš„çµŒç·¯ã‚’è€ƒãˆã‚‹ã¨å¼·åŒ–å¦ç¿’+æ·±å±¤å¦ç¿’ã«ãªã£ãŸã€ã‚€ã—ã‚ã‹ãªã‚Šå½“ç„¶ã®æˆã‚Šè¡Œãã§ç”Ÿã¾ã‚ŒãŸæŠ€è¡“ã§ã‚ã‚‹ã“ã¨ãŒã‚ã‹ã‚Šã¾ã™ã€‚ï¼ˆATARIã®ã‚²ãƒ¼ãƒ ã‚’äººé–“ä»¥ä¸Šã®ãƒ‘ãƒ•ã‚©ãƒ¼ãƒžãƒ³ã‚¹ã§ãƒ—ãƒ¬ã‚¤ã™ã‚‹ã¨ã„ã†ã®ãŒãƒ“ã‚¸ãƒ¥ã‚¢ãƒ«çš„ã«ã‚ã‹ã‚Šã‚„ã™ã‹ã£ãŸ$\leftrightarrow$å•é¡Œè¨å®šãŒè‰¯ã‹ã£ãŸã¨ã„ã†ã®ã‚‚ã‚ã‚Šã¾ã™ã€‚ï¼‰ ã“ã®è¨˜äº‹ã§ã¯NIPSã¨Natureã®ä»¥ä¸‹ã®ï¼’æœ¬ã®è«–æ–‡ ãƒ»
p_tan 2015/07/14
deep learning

æ©Ÿæ¢°å¦ç¿’

å¼·åŒ–å¦ç¿’

machine learning

chainer

deep_learning

DQN
ãƒªãƒ³ã‚¯
chainerã§ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆã‚’å¦ã‚“ã§ã¿ã‚‹ã‚ˆ(chainerã§ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆ2) - äººå·¥è¨€èªžå‡¦ç†å…¥é–€
å‰å›žã®è¨˜äº‹ã§ã¯chainerã®ã‚¤ãƒ³ã‚¹ãƒˆãƒ¼ãƒ«ã‹ã‚‰ã‚µãƒ³ãƒ—ãƒ«ã‚³ãƒ¼ãƒ‰ã‚’ä½¿ã£ã¦ç”»åƒè˜åˆ¥å•é¡Œã‚’è§£ãã¨ã“ã‚ã¾ã§ã‚„ã‚Šã¾ã—ãŸã€‚ hi-king.hatena blog.com ä»Šå›žã®è¨˜äº‹ã§ã¯å›žå¸°ãƒ»åˆ†é¡žå•é¡Œç”¨ã®ã‚·ãƒ³ãƒ—ãƒ«ãªãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆã®ä½œã‚Šæ–¹ã‚’ã‚„ã‚ã†ã¨æ€ã„ã¾ã™ã€‚andã¨xorã®è«–ç†å¼ã‚’å¦ç¿’ã•ã›ã¾ã™ã€‚chainerã§ã®å®Ÿè£…ã®å¦ç¿’ã¨ã€ã‚ã¨ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆã®æ•™è‚²ç›®çš„ã«ä½¿ã„ã‚„ã™ã„ãªãƒ¼ã¨æ€ã£ãŸã®ã§ã€‚2å±¤ã®ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆã¾ã§æ®µéšŽã‚’ãµã‚“ã§è§£èª¬ã—ã¦ã‚‹ã‚“ã§ã™ãŒã€ãƒ—ãƒã‚°ãƒ©ãƒ èªã‚€ã»ã†ãŒå¾—æ„ã€ã£ã¦æ–¹ã¯ä¸€ç•ªæœ€å¾Œã®ã‚³ãƒ¼ãƒ‰ã‚’å…ˆã«èªã‚“ã ã»ã†ãŒã‚ã‹ã‚Šã‚„ã™ã„ã‹ã‚‚ã—ã‚Œã¾ã›ã‚“ã€‚ è¿½è¨˜(7/13)åž‹ãƒã‚§ãƒƒã‚¯ chainer1.1.0ã‹ã‚‰åž‹ãƒã‚§ãƒƒã‚¯ãŒå…¥ã£ãŸã®ã§(https://github.com/pfnet/chainer/pull/95)ã€è˜åˆ¥ã«ã¯float32ã‚’å…¥åŠ›ã—ã¦int32ã‚’å‡ºåŠ›ã€å›žå¸°ã«ã¯float32ã‚’å…¥åŠ›ã—ã¦float32ã‚’å‡ºåŠ›ã€ã¨ã„
p_tan 2015/06/28
chainer

python

deep learning

MachineLearning

machine-learning
ãƒªãƒ³ã‚¯
1