[B! neural-network] stibbarã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯

stibbar id:stibbar

neural-networkã«é–¢ã™ã‚‹stibbarã®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}}

èª¤å·®é€†ä¼æ’æ³•ã‚’ã¯ã˜ã‚ã‹ã‚‰ã¦ã„ãã„ã« - Qiita
ã¯ã˜ã‚ã« ã“ã®è¨˜äº‹ã¯ã€ç§ãŒæ©Ÿæ¢°å¦ç¿’ã€ç‰¹ã«ã€Œèª¤å·®é€†ä¼æ’æ³•ã€ (ã‚ã‚‹ã„ã¯ã€Œãƒãƒƒã‚¯ãƒ—ãƒãƒ‘ã‚²ãƒ¼ã‚·ãƒ§ãƒ³ (Backpropagation)ã€) ã‚’å‹‰å¼·ã™ã‚‹ã†ãˆã§è¾¿ã£ãŸé“ç‹ã‚’è¨˜éŒ²ã—ã¦ãŠãã‚‚ã®ã§ã™ã€‚ èª¤å·®é€†ä¼æ’æ³•ã«é–¢ã™ã‚‹åˆ†ã‹ã‚Šã‚„ã™ã„è§£èª¬ã¯ã€ã“ã“ Qiita ã«ã‚‚å¤šæ•°å˜åœ¨ã—ã¾ã™ã®ã§ã€æ£ç›´ $n$ ç•ªç…Žã˜ãªã¨ã“ã‚ãŒã‚ã‚Šã¾ã™ãŒã€ã‚ãã¾ã§è¨˜éŒ²ã§ã™ã®ã§ã“ã“ã¯ 1 ã¤ã€‚ ã€Œè¨˜éŒ²ã€ã¨ã„ã†ã«ã¯é•·éŽãŽã¦ã‚¤ãƒžã‚¤ãƒæ„Ÿæº¢ã‚Œã¦ã¾ã™ãŒã€ãã“ã¯æ°—ã«ã—ãªã„ã§ãã ã•ã„ã€‚ã“ã‚Œã§ã‚‚è¦ç‚¹ã‚’ã—ã¼ã£ãŸã¤ã‚‚ã‚Šãªã®ã§ã™â€¦ ã‚†ã£ãã‚Šã—ã¦ã„ã£ã¦ãï¼ï¼ï¼ é€”ä¸ã®æ•°å¼ã«ã¤ã„ã¦ è¨˜äº‹ã®å‰å¾Œã‹ã‚‰å‚ç…§ã•ã‚Œã¦ã„ã‚‹æ•°å¼ã«ã¤ã„ã¦ã¯å³å´ã«ç•ªå·ã‚’æŒ¯ã£ã¦ã„ã¾ã™ã€‚ åˆå›žè¨˜è¼‰æ™‚ã¯ $\text{(1)}$ ã¨ã‹ã£ã“ä»˜ãæ•°å—ã§ã€2 å›žç›®ä»¥é™è¨˜è¼‰æ™‚ã¯ $\text{[1]}$ ã¨è§’ã‹ã£ã“ä»˜ãæ•°å—ã§ç¤ºã—ã¾ã™ã€‚è¨˜è¼‰æ¸ˆã¿ã®æ•°å¼ã‚’å¤‰å½¢ã—ãŸã‚‚ã®ã«ã¤ã„ã¦ã¯ $\text{(1')}$ ã¨ãƒ—ãƒ©ã‚¤ãƒ
stibbar 2018/06/20
http://d.hatena.ne.jp/nowokay/20080701/1214915017

qiita

neural-network

back-propagation

machine-learning
ãƒªãƒ³ã‚¯
æ·±å±¤å¦ç¿’ã®ãƒ¢ãƒ‡ãƒ«åœ§ç¸®ãƒ»é«˜é€ŸåŒ–ã«é–¢ã™ã‚‹è«–æ–‡80æœ¬ãƒŽãƒƒã‚¯ - ãã†ã ã
ã“ã‚“ã«ã¡ã¯.ç„¡äº‹ã«3å¹´ç”Ÿã«é€²ç´šã§ãã¦ãƒ›ãƒƒã¨ã—ã¦ã„ã¾ã™. æ˜¨å¹´åº¦ã‹ã‚‰æ·±å±¤å¦ç¿’ã®ãƒ¢ãƒ‡ãƒ«æ”¹å¤‰ã«ã‚ˆã‚‹é«˜é€ŸåŒ–æ‰‹æ³•ã«ã¤ã„ã¦èª¿ã¹ã¦ã„ã¾ã—ãŸ.ãã‚“ãªã‚ã‘ã§æœ¬ç¨¿ã§ã¯æ·±å±¤å¦ç¿’ãƒ¢ãƒ‡ãƒ«ã®ãƒ‘ãƒ©ãƒ¡ãƒ¼ã‚¿å®¹é‡ã®å‰Šæ¸›ã‚„é«˜é€ŸåŒ–ã‚’ç›®çš„ã¨ã—ãŸæ‰‹æ³•ã«é–¢ã™ã‚‹è«–æ–‡ã‚’80æœ¬ç´¹ä»‹ã—ã¾ã™. ä¸€éƒ¨, ãƒã‚¤ãƒ‘ãƒ¼ãƒ‘ãƒ©ãƒ¡ãƒ¼ã‚¿æŽ¢ç´¢(Neural Architecture Searchç‰)ã®è«–æ–‡ã‚’ã€Œé«˜é€Ÿãªãƒ¢ãƒ‡ãƒ«ã®æŽ¢ç´¢ã«ä½¿ãˆãªã„ã‹ãªã€œã€ã¨ã„ã†ãã‚‚ã¡ã§å«ã‚ã¾ã—ãŸ. ã¾ãŸ, è’¸ç•™ã«é–¢ã™ã‚‹è«–æ–‡ã¯å¿…ãšã—ã‚‚é«˜é€ŸåŒ–ã‚’ç‹™ã£ãŸã‚‚ã®ã ã‘ã§ã¯ãªã, è’¸ç•™ãã®ã‚‚ã®ã®æ´¾ç”Ÿè«–æ–‡ã«ã¤ã„ã¦ã‚‚å«ã‚ã¦ã„ã¾ã™. å…¨ã¦ã®è«–æ–‡ã«ã¤ã„ã¦è©³è§£ã§ã¯ãªãé›‘ãªæ¦‚è¦ã®ã¿è¨˜è¼‰ã—ã¦ã„ã¾ã™. ã‚µãƒ¼ãƒ™ã‚¤ã™ã‚‹éš›ã®ã‚¬ã‚¤ãƒ‰ã«ã§ã‚‚ä½¿ã£ã¦ã„ãŸã ã‘ã‚Œã°å¹¸ã„ã§ã™. ã‚µã‚¯ãƒƒã¨èªã‚“ã ã ã‘ã®è«–æ–‡ã‚‚å¤šã€…ã‚ã‚Šã¾ã™ã®ã§é–“é•ã„ç‰ã”ã–ã„ã¾ã—ãŸã‚‰ã”æŒ‡æ‘˜ãã ã•ã„. ã§ã¯å‚ã‚Šã¾ã—ã‚‡ã†. é«˜é€Ÿãªãƒ¢ãƒ‡ãƒ«ã‚¢ãƒ¼ã‚ãƒ†ã‚¯ãƒãƒ£ ã‚ˆã‚Šè‰¯ã„é€Ÿåº¦ã¨ç²¾åº¦ã®ãƒˆãƒ¬ãƒ¼ãƒ‰ã‚ªãƒ•ã‚’
stibbar 2018/06/16
machine-learning

deep-learning

neural-network

distillation
ãƒªãƒ³ã‚¯
ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ - Wikipedia
ï¼ˆäººå·¥çŸ¥èƒ½ã®åˆ†é‡Žã§ï¼‰ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ï¼ˆè‹±: neural network; NNã€ç¥žçµŒç¶²ï¼‰ã¯ã€ç”Ÿç‰©ã®å¦ç¿’ãƒ¡ã‚«ãƒ‹ã‚ºãƒ ã‚’æ¨¡å€£ã—ãŸæ©Ÿæ¢°å¦ç¿’æ‰‹æ³•ã¨ã—ã¦åºƒãçŸ¥ã‚‰ã‚Œã¦ã„ã‚‹ã‚‚ã®ã§ã‚ã‚Š[1]ã€ã€Œãƒ‹ãƒ¥ãƒ¼ãƒãƒ³ã€ã¨å‘¼ã°ã‚Œã‚‹è¨ˆç®—ãƒ¦ãƒ‹ãƒƒãƒˆã‚’ã‚‚ã¡ã€ç”Ÿç‰©ã®ç¥žçµŒç³»ã®ãƒ¡ã‚«ãƒ‹ã‚ºãƒ ã‚’æ¨¡å€£ã—ã¦ã„ã‚‹ã‚‚ã®ã§ã‚ã‚‹[1]ã€‚äººé–“ã®è„³ã®ç¥žçµŒç¶²ã‚’æ¨¡ã—ãŸæ•°ç†ãƒ¢ãƒ‡ãƒ«[2]ã€‚æ¨¡å€£å¯¾è±¡ã¨ãªã£ãŸç”Ÿç‰©ã®ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ï¼ˆç¥žçµŒç¶²ï¼‰ã¨ã¯ã£ãã‚ŠåŒºåˆ¥ã™ã‚‹å ´åˆã¯ã€äººå·¥ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ (è‹±: artificial neural network) ã¨å‘¼ã°ã‚Œã‚‹ã€‚ ä»¥ä¸‹ã§ã¯èª¬æ˜Žã®éƒ½åˆä¸Š[æ³¨é‡ˆ 1]ã€äººå·¥çš„ãªãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã®ã»ã†ã¯ã€Œäººå·¥ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã€ã‚ã‚‹ã„ã¯å˜ã«ã€Œãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã€ã¨å‘¼ã³ã€ç”Ÿç‰©ã®ãã‚Œã¯ã€Œç”Ÿç‰©ã®ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã€ã‚ã‚‹ã„ã¯ã€Œç”Ÿç‰©ã®ç¥žçµŒç¶²ã€ã€ãƒ’ãƒˆã®é è„³ã®ãã‚Œã¯ã€Œãƒ’ãƒˆã®ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã€ã‚ã‚‹ã„ã¯ã€Œãƒ’ãƒˆã®ç¥žçµŒç¶²ã€
stibbar 2018/06/13
FFNN

Neural-network

wikipedia
ãƒªãƒ³ã‚¯
æœ€è¿‘ã®ã‚¹ãƒ‘ãƒ¼ã‚¹ãªãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã«ã¤ã„ã¦ - SmartNews Engineering Blog
ã“ã‚“ã«ã¡ã¯ã€ã‚¹ãƒžãƒ¼ãƒˆãƒ‹ãƒ¥ãƒ¼ã‚¹ã®å¾³æ°¸ã§ã™ã€‚æ·±å±¤å¦ç¿’æ¥ç•Œã¯GANã ã¨ã‹æ·±å±¤å¼·åŒ–å¦ç¿’ã ã¨ã‹ã§ç››ã‚Šä¸ŠãŒã£ã¦ã„ã¾ã™ãŒã€ä»Šæ—¥ã¯æ·¡ã€…ã¨ã€ã‚¹ãƒ‘ãƒ¼ã‚¹ãªãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã®è©±ã‚’ã—ã¾ã™ã€‚ è¦ç´„ã™ã‚‹ã¨ ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆã®ã‚¹ãƒ‘ãƒ¼ã‚¹åŒ–ã«ã‚ˆã£ã¦ã€ç²¾åº¦ã¯ã»ã¨ã‚“ã©çŠ ç‰²ã«ã›ãšã«ã€è¨ˆç®—åŠ¹çŽ‡ã¯3ã€œ5å€ç¨‹åº¦ã¾ã§å‘ä¸Šã§ãã‚‹ ã‚¹ãƒ‘ãƒ¼ã‚¹åŒ–ã«ã¯ã¾ã èª²é¡ŒãŒå¤šãã€ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆã®é«˜é€ŸåŒ–ã¨ã„ã†æ„å‘³ã§ã¯ã€æ¬¡ã®æˆ¦å ´ã¯ãŸã¶ã‚“ã“ã“ã«ãªã‚‹ ã‚¹ãƒ‘ãƒ¼ã‚¹ã¨ã¯ã€ã‚¹ãƒ‘ãƒ¼ã‚¹åŒ–ã¨ã¯ ã‚¹ãƒ‘ãƒ¼ã‚¹ã§ã‚ã‚‹ã¨ã¯ã€å€¤ã®ã»ã¨ã‚“ã©ãŒ0ã§ã‚ã‚‹ã“ã¨ã§ã™ã€‚ä¾‹ãˆã°ã€ãƒ™ã‚¯ãƒˆãƒ«$a,b$ã®å†…ç©ã‚’è¨ˆç®—ã™ã‚‹éš›ã«ã€$a$ã®ã»ã¨ã‚“ã©ã®è¦ç´ ã®å€¤ãŒ0ã§ã‚ã‚‹ã¨ã—ã¾ã—ã‚‡ã†ã€‚0ã«ãªã«ã‚’ã‹ã‘ã¦ã‚‚0ã§ã™ã‹ã‚‰ã€$a$ã®å€¤ãŒ0ã§ãªã„æ¬¡å…ƒã«ã¤ã„ã¦ã®ã¿ã€$a_i b_i$ã®å€¤ã‚’è¨ˆç®—ã—ã¦è¶³ã—åˆã‚ã›ã‚Œã°ã‚ˆã„ã‚ã‘ã§ã™ã€‚ã“ã®ã‚ˆã†ã«ã€å†…ç©ã‚’è¨ˆç®—ã™ã‚‹éš›ã«ã€ã©ã¡ã‚‰ã‹ã®ãƒ™ã‚¯ãƒˆãƒ«ãŒã‚¹ãƒ‘ãƒ¼ã‚¹ã§ã‚ã‚Œã°è¨ˆç®—ãŒé«˜é€ŸåŒ–ã§ãã¾ã™ã€‚0ã¨ã„ã†å€¤ã‚’ãƒ¡ãƒ¢ãƒª
stibbar 2018/05/31
https://www.sciencedirect.com/science/article/pii/S0031320315001971

sparse

smartnews

neural-network
ãƒªãƒ³ã‚¯
Scott Rome
stibbar 2018/03/13
deep-learning

neural-network

machine-learning

reinforcement-learning
ãƒªãƒ³ã‚¯
Neural Style Transfer: Prismaã®èƒŒæ™¯æŠ€è¡“ã‚’è§£èª¬ã™ã‚‹
å†™çœŸã‚’ãƒ”ã‚«ã‚½ã‚„ã‚´ãƒƒãƒ›ã®ã‚ˆã†ãªã‚¹ã‚¿ã‚¤ãƒ«ã«å¤‰æ›ã§ãã‚‹ã‚¢ãƒ—ãƒªPrismaãŒè©±é¡Œã«ãªã‚Šã¾ã—ãŸã€‚å¤šãã®äººã¯ã€ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ãŒä½¿ã‚ã‚Œã¦ã„ã‚‹ã‹ã©ã†ã‹ã¨ã¯é–¢ä¿‚ãªãã€ç´”ç²‹ã«ã‚¢ãƒ—ãƒªã‚’æ¥½ã—ã‚“ã§ã„ã‚‹ã®ã ã¨æ€ã„ã¾ã™ã€‚ ã“ã®ã‚ˆã†ã«ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ã‚’ä½¿ã£ãŸäººæ°—ã‚¢ãƒ—ãƒªãŒå‡ºã¦ãã‚‹ã¨ã„ã†ã“ã¨ã¯éžå¸¸ã«è‰¯ã„ã“ã¨ã§ã¯ãªã„ã‹ã¨æ€ã„ã¾ã™ã€‚ä»Šå›žã¯ã€Prismaã®èƒŒæ™¯æŠ€è¡“ï¼ˆã¨æ€ã‚ã‚Œã‚‹ã‚‚ã®ï¼‰ã‚’è§£èª¬ã—ã¦ã„ãã¾ã™ã€‚ ç›®æ¬¡ åŸºç¤Žç†è«– å®Ÿè£… æ”¹å–„ é«˜é€ŸåŒ– ã¾ã¨ã‚ åŸºç¤Žç†è«– ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ã‚’ä½¿ã£ãŸã‚¢ãƒ¼ãƒˆç³»ã®è«–æ–‡ã¯è‰²ã€…ã¨å‡ºã¦ã„ã¾ã™ãŒã€ä¸€ç•ªåŸºç¤Žã¨ãªã‚‹è«–æ–‡ã¯Gatys et al. 2016ã§ã¯ãªã„ã‹ã¨æ€ã„ã¾ã™ã€‚ãƒ—ãƒ¬ãƒ—ãƒªãƒ³ãƒˆç‰ˆã¯2015å¹´8æœˆã«å‡ºã¦ã„ã¾ã™ã€‚ ã“ã®è«–æ–‡ã¯è¨˜äº‹ã¨ã—ã¦å–ã‚Šä¸Šã’ã‚‰ã‚Œã¦è©±é¡Œã«ãªã£ã¦ã„ãŸã‚Šã‚‚ã—ãŸã®ã§ã€çŸ¥ã£ã¦ã„ã‚‹äººã‚‚å¤šã„ã®ã§ã¯ãªã„ã‹ã¨æ€ã„ã¾ã™ã€‚ã“ã®ç« ã§ã¯ã€ã‚¹ã‚¿ã‚¤ãƒ«å¤‰æ›ã®åŸºç¤Žã¨ãªã‚‹ã“ã®è«–æ–‡ã‚’è§£èª¬ã—ã¦ã„ãã¾ã™ã€‚ Gatys et a
stibbar 2018/02/25
deep-learning

neural-network
ãƒªãƒ³ã‚¯
ã€Žæ±å¤§ã‚‰ã€ç¡çœ ä¸ã«è„³ãŒã‚¯ãƒ¼ãƒ«ãƒ€ã‚¦ãƒ³ã•ã‚Œã‚‹ä»•çµ„ã¿ã‚’è§£æ˜Ž- ç”Ÿç‰©ãŒçœ ã‚‹è¬Žã«è¿«ã‚‹ã€ã¸ã®ã‚³ãƒ¡ãƒ³ãƒˆ
ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ã—ã¾ã—ãŸ ã“ã“ã«ãƒ„ã‚¤ãƒ¼ãƒˆå†…å®¹ãŒè¨˜è¼‰ã•ã‚Œã¾ã™ https://b.hatena.ne.jp/URLã¯spanã§å›²ã‚“ã§ãã ã•ã„ Twitterã§å…±æœ‰
stibbar 2018/02/23
neural-network

neurology

UT

hatena-bookmark
ãƒªãƒ³ã‚¯
Word2Vecï¼šç™ºæ˜Žã—ãŸæœ¬äººã‚‚é©šãå˜èªžãƒ™ã‚¯ãƒˆãƒ«ã®é©šç•°çš„ãªåŠ›
Word2Vecã¨ã¯ Word2Vecã§æ¼”ç®—å‡¦ç†ã™ã‚‹ Word2Vecã¨ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ Word2Vecã®ä»•çµ„ã¿ CBoW Skip-gram Word2Vecã‚’å¿œç”¨ã™ã‚‹ã“ã¨ãŒã§ãã‚‹åˆ†é‡Ž ãƒ¬ã‚³ãƒ¡ãƒ³ãƒ‰ æ©Ÿæ¢°ç¿»è¨³ Q&Aãƒ»ãƒãƒ£ãƒƒãƒˆãƒœãƒƒãƒˆ æ„Ÿæƒ…åˆ†æž Word2Vecã®å¼±ç‚¹ Word2Vecã®æ´¾ç”Ÿç³»ã‚„é¡žä¼¼ãƒ„ãƒ¼ãƒ« GloVe WordNet Doc2Vec fastText ã¾ã¨ã‚ å‚è€ƒ ä¸–ç•Œä¸ã®Webã‚µã‚¤ãƒˆã®æ•°ã¯2014å¹´ã«10å„„ä»¶ã‚’è¶…ãˆãŸã‚ˆã†ã ã€‚ãã—ã¦ã€Facebookã®ãƒ¦ãƒ¼ã‚¶ãƒ¼æ•°ã ã‘ã§ã‚‚16å„„äººã‚’è¶…ãˆã¦ã„ã‚‹ã€‚ ãã—ã¦ã€ãã®ã„ãšã‚Œã‚‚ã‚³ãƒ³ãƒ†ãƒ³ãƒ„ã®ä¸èº«ã®å¤§éƒ¨åˆ†ã¯ãƒ†ã‚ã‚¹ãƒˆã‹ã‚‰æˆã‚Šç«‹ã£ã¦ã„ã‚‹ã“ã¨ã ã‚ã†ã€‚ ã¨ã„ã†ã“ã¨ã¯ã€èŽ«å¤§ã«å¢—å¤§ã—ç¶šã‘ã‚‹ãƒãƒƒãƒˆä¸Šã®ãƒ‡ãƒ¼ã‚¿ã®ã»ã¨ã‚“ã©ã¯ã©ã“ã‹ã®å›½ã®è¨€è‘‰ã ã£ã¦ã“ã¨ã ã€‚ä¸–ç•Œä¸ã®äººãŒæ¯Žæ—¥ãƒ†ã‚ã‚¹ãƒˆãƒ‡ãƒ¼ã‚¿ã‚’ç”Ÿæˆã—ç¶šã‘ãŸã“ã¨ã¯ã“ã‚Œã¾ã§ã®æ´å²ä¸Šç„¡ã‹ã£ãŸã‚“ã˜ã‚ƒãªã„ã ã‚ã†ã‹ã€‚ ã‚‚ã—ãã‚“
stibbar 2018/02/17
machine-learning

neural-network
ãƒªãƒ³ã‚¯
é£¯å¡šé‡Œå¿— â€” ãƒ‡ã‚£ãƒ¼ãƒ—ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã‚’ç”¨ã„ãŸå¤§åŸŸç‰¹å¾´ã¨å±€æ‰€ç‰¹å¾´ã®å¦ç¿’ã«ã‚ˆã‚‹ç™½é»’å†™çœŸã®è‡ªå‹•è‰²ä»˜ã‘
æ¦‚è¦ï¼š æœ¬ç ”ç©¶ã§ã¯ï¼Œãƒ‡ã‚£ãƒ¼ãƒ—ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã‚’ç”¨ã„ã¦ç™½é»’ç”»åƒã‚’ã‚«ãƒ©ãƒ¼ç”»åƒã«è‡ªå‹•å¤‰æ›ã™ã‚‹æ‰‹æ³•ã‚’ææ¡ˆã™ã‚‹ï¼Žææ¡ˆæ‰‹æ³•ã§ã¯ï¼Œç”»åƒã®å¤§åŸŸç‰¹å¾´ã¨å±€æ‰€ç‰¹å¾´ã‚’è€ƒæ…®ã—ãŸæ–°ãŸãªç•³è¾¼ã¿ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ãƒ¢ãƒ‡ãƒ«ã‚’ç”¨ã„ã‚‹ã“ã¨ã§ï¼Œç”»åƒå…¨ä½“ã®æ§‹é€ ã‚’è€ƒæ…®ã—ãŸè‡ªç„¶ãªè‰²ä»˜ã‘ã‚’è¡Œã†ã“ã¨ãŒã§ãã‚‹ï¼Žææ¡ˆãƒ¢ãƒ‡ãƒ«ã«ãŠã„ã¦ï¼Œå¤§åŸŸç‰¹å¾´ã¯ç”»åƒå…¨ä½“ã‹ã‚‰æŠ½å‡ºã•ã‚Œï¼Œå±€æ‰€ç‰¹å¾´ã¯ã‚ˆã‚Šå°ã•ãªç”»åƒé ˜åŸŸã‹ã‚‰è¨ˆç®—ã•ã‚Œã‚‹ï¼Žã“ã‚Œã‚‰ã®ç‰¹å¾´ã¯â€œçµåˆãƒ¬ã‚¤ãƒ¤â€ã«ã‚ˆã£ã¦ä¸€ã¤ã«çµ±åˆã•ã‚Œï¼Œè‰²ä»˜ã‘ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã«å…¥åŠ›ã•ã‚Œã‚‹ï¼Žã“ã®ãƒ¢ãƒ‡ãƒ«æ§‹é€ ã¯å…¥åŠ›ç”»åƒã®ã‚µã‚¤ã‚ºãŒå›ºå®šã•ã‚Œãšï¼Œã©ã‚“ãªã‚µã‚¤ã‚ºã®ç”»åƒã§ã‚‚å…¥åŠ›ã¨ã—ã¦ç”¨ã„ã‚‹ã“ã¨ãŒã§ãã‚‹ï¼Žã¾ãŸï¼Œãƒ¢ãƒ‡ãƒ«ã®å¦ç¿’ã®ãŸã‚ã«æ—¢å˜ã®å¤§è¦æ¨¡ãªç”»åƒåˆ†é¡žã®ãƒ‡ãƒ¼ã‚¿ã‚»ãƒƒãƒˆã‚’åˆ©ç”¨ã—ï¼Œãã‚Œãžã‚Œã®ç”»åƒã®è‰²ã¨ãƒ©ãƒ™ãƒ«ã‚’åŒæ™‚ã«å¦ç¿’ã«ç”¨ã„ã‚‹ã“ã¨ã§ï¼ŒåŠ¹æžœçš„ã«å¤§åŸŸç‰¹å¾´ã‚’å¦ç¿’ã§ãã‚‹ã‚ˆã†ã«ã—ã¦ã„ã‚‹ï¼Žææ¡ˆæ‰‹æ³•ã«ã‚ˆã‚Šï¼Œ100å¹´å‰ã®ç™½é»’å†™çœŸãªã©ï¼Œæ§˜ã€…ãªç”»åƒã«ãŠã„ã¦è‡ªç„¶ãªè‰²ä»˜ã‘ã‚’å®Ÿç¾ã§ãã‚‹ï¼Žè‰²ä»˜ã‘ã®çµæžœã¯ãƒ¦ãƒ¼
stibbar 2016/05/09
deep-network

neural-network

image

ML
ãƒªãƒ³ã‚¯
ãƒ‡ã‚£ãƒ¼ãƒ—ãƒ©ãƒ¼ãƒ‹ãƒ³ã‚°ï¼ç•³ã¿è¾¼ã¿ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã¨Pythonã«ã‚ˆã‚‹ç‰¹å¾´æŠ½å‡º | POSTD
stibbar 2015/10/30
deep-learning

python

neural-network
ãƒªãƒ³ã‚¯
1