[B! Programming][æ©Ÿæ¢°å¦ç¿’] ihokã®ãƒ

ihok id:ihok

Programmingã¨æ©Ÿæ¢°å¦ç¿’ã«é–¢ã™ã‚‹ihokã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ (5)

${{author_name}}$

{{{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}}

ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã®é‡ååŒ–ã«ã¤ã„ã¦ã®æœ€è¿‘ã®ç ”ç©¶ã®é€²å±•ã¨ã€ãã®é‡è¦æ€§ - 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€ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã®é‡ååŒ–ã®è©±ã‚’ã—ã¾ã™ã€‚ TL;DR ãƒ‘ãƒ©ãƒ¡ãƒ¼ã‚¿ã ã‘ã‚’é‡ååŒ–ã™ã‚‹ã®ã§ã‚ã‚Œã°ã€ã»ã¼ç²¾åº¦ã‚’è½ã¨ã•ãšã«ã€ãƒ‘ãƒ©ãƒ¡ãƒ¼ã‚¿ã®ãƒ‡ãƒ¼ã‚¿å®¹é‡ã¯1/16ç¨‹åº¦ã«ã¾ã§å‰Šæ¸›ã§ãã‚‹ ãƒ‘ãƒ©ãƒ¡ãƒ¼ã‚¿ã€ã‚¢ã‚¯ãƒ†ã‚£ãƒ™ãƒ¼ã‚·ãƒ§ãƒ³ã€å‹¾é…ã®ã™ã¹ã¦ã‚’é‡ååŒ–ã—ã€æŽ¨è«–ã ã‘ã§ãªãå¦ç¿’ã¾ã§ã‚‚ãƒ“ãƒƒãƒˆæ¼”ç®—ã ã‘ã§å®Ÿç¾ã™ã‚‹ç ”ç©¶ãŒé€²ã‚“ã§ãã¦ã„ã‚‹ ç¾åœ¨ã¯æ·±å±¤å¦ç¿’ = GPUå¿…é ˆã¨ã„ã†ãã‚‰ã„ã®å‹¢ã„ãŒã‚ã‚‹ã‘ã©ã€é‡ååŒ–ã®ç ”ç©¶ãŒé€²ã‚€ã¨ã€ä»Šå¾Œã¯ã©ã†ãªã‚‹ã‹ã‚ã‹ã‚‰ãªã„ã‚ˆ ã¯ã˜ã‚ã« æƒ…å ±ç†è«–ã«ãŠã‘ã‚‹é‡ååŒ–ã¨ã¯ã€ã‚¢ãƒŠãƒã‚°ãªé‡ã‚’é›¢æ•£çš„ãªå€¤ã§è¿‘ä¼¼çš„ã«è¡¨ç¾ã™ã‚‹ã“ã¨ã‚’æŒ‡ã—ã¾ã™ãŒã€æœ¬ç¨¿ã«ãŠã‘ã‚‹é‡ååŒ–ã¯åŽ³å¯†ã«è¨€ã†ã¨ã¡ã‚‡ã£ã¨æ„å‘³ãŒé•ã„ã€ååˆ†ãªï¼ˆ=32bitã‚‚ã—ãã¯16bitï¼‰ç²¾åº¦ã§è¡¨ç¾ã•ã‚Œã¦ã„ãŸé‡ã‚’ã€ãšã£ã¨å°‘ãªã„ãƒ“ãƒƒãƒˆæ•°ã§è¡¨ç¾ã™ã‚‹ã“ã¨ã‚’è¨€ã„ã¾ã™ã€‚ ãƒ‹ãƒ¥
ihok 2017/03/08
gpu

AI

æ©Ÿæ¢°å¦ç¿’

machinelearning

ç ”ç©¶

NLP

programming

performance

ã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ
ãƒªãƒ³ã‚¯
æ–‡ç« ã‚’ãƒ™ã‚¯ãƒˆãƒ«åŒ–ã—ã¦é¡žä¼¼æ–‡ç« ã®æ¤œç´¢ - Qiita
Doc2Vecã§é¡žä¼¼æ–‡ç« ã‚’æ¤œç´¢ã—ã¦ã¿ãŸã®ã§ã€å®Ÿè£…ã‚’ç´¹ä»‹ã—ã¾ã™ã€‚ Doc2Vecã¨ã¯ ã‚³ãƒ³ãƒ”ãƒ¥ãƒ¼ã‚¿ãŒè‡ªç„¶è¨€èªžã‚’å‡¦ç†ã™ã‚‹ãŸã‚ã«ã¯ã€ã¾ãšäººé–“ã®è¨€è‘‰ã‚’ã‚³ãƒ³ãƒ”ãƒ¥ãƒ¼ã‚¿ã§æ‰±ãˆã‚‹å€¤ã«ã™ã‚‹å¿…è¦ãŒã‚ã‚Šã¾ã™ã€‚ å˜èªžã®æ„å‘³ã‚’ãƒ™ã‚¯ãƒˆãƒ«åŒ–ã™ã‚‹æ‰‹æ³•ã¨ã—ã¦Word2VecãŒå˜åœ¨ã—ã¾ã™ã€‚ è©³ã—ãã¯ãƒªãƒ³ã‚¯å…ˆãŒã¨ã¦ã‚‚ã‚ã‹ã‚Šã‚„ã™ã„ã®ã§ã™ãŒã€ã–ã£ãã‚Šè¨€ã†ã¨å‰å¾Œnå˜èªžã®ãƒªã‚¹ãƒˆã§ãã®å˜èªžã‚’è¡¨ç¾ã—ã¾ã™ã€‚ ã“ã†ã™ã‚‹ã“ã¨ã§ã€ä¾‹ãˆã°ã€ŒçŠ¬ã€ã¨ã€ŒçŒ«ã€ã¯åŒã˜ã‚ˆã†ãªæ–‡è„ˆã§ä½¿ã‚ã‚Œã‚‹ãŸã‚ã€ä¼¼ãŸã€Œæ„å‘³ã€ã§ã‚ã‚‹ã¨è€ƒãˆã‚‹ã“ã¨ãŒã§ãã¾ã™ã€‚ Doc2Vecã¯Word2Vecã‚’å¿œç”¨ã—ã€æ–‡ç« ã‚’ãƒ™ã‚¯ãƒˆãƒ«åŒ–ã™ã‚‹ã‚‚ã®ã§ã™ã€‚ å®Ÿè£…ã‚µãƒ³ãƒ—ãƒ« ä»Šå›žDoc2Vecã‚’ç”¨ã„ã¦å®Ÿç¾ã™ã‚‹ã®ã¯ã€ä»¥ä¸‹ã®2ã¤ã®æ©Ÿèƒ½ã§ã™ã€‚ å˜èªžã§æ–‡ç« ã‚’æ¤œç´¢ é¡žä¼¼æ–‡ç« ã®æ¤œç´¢ ã‚µãƒ³ãƒ—ãƒ«ã¨ã—ã¦ã€é’ç©ºæ–‡åº«ã®æ–‡ç« ã‚’ä½¿ç”¨ã—ã¾ã—ãŸã€‚ ãªãŠã€ã“ã®è¨˜äº‹ã§ä½¿ç”¨ã™ã‚‹ã‚³ãƒ¼ãƒ‰ã¯GitHubã§å…¬é–‹ã—ã¦ã„ã¾ã™ã€‚ (å¦ç¿’ã«ä½¿ç”¨ã—ãŸæ–‡ç« ã‚‚zipã«
ihok 2017/02/28
github

word2vec

ãƒ™ã‚¯ãƒˆãƒ«

ä»•äº‹

python

ã‚³ãƒ³ãƒ”ãƒ¥ãƒ¼ã‚¿

æ©Ÿæ¢°å¦ç¿’

programming
ãƒªãƒ³ã‚¯
æ©Ÿæ¢°å¦ç¿’ã‚’ãƒ—ãƒãƒ€ã‚¯ãƒˆã«å…¥ã‚Œã‚‹éš›ã«è€ƒãˆã‚‹æŽ¡ç”¨åŸºæº–ã«ã¤ã„ã¦ - yasuhisa's blog
ã‚µãƒ¼ãƒ“ã‚¹ã«æ©Ÿæ¢°å¦ç¿’æŠ€è¡“(ä¾‹ãˆã°SVM)ã‚’å…¥ã‚Œã‚‹éš›ã«ã€ã€Œã“ã®æ©Ÿæ¢°å¦ç¿’æŠ€è¡“ã¯æœ¬ç•ªã‚µãƒ¼ãƒ“ã‚¹ã«æŠ•å…¥ã—ã¦ã‚‚å¤§ä¸ˆå¤«ãªã‚‚ã®ã‹?ã€ã‚’è€ƒãˆã‚‹åŸºæº–ãŒã¾ã¨ã¾ã£ã¦ã„ã‚‹ã¨äººã«èª¬æ˜Žã™ã‚‹ã¨ãã«ä¾¿åˆ©ã ãªã¨ãµã¨æ€ã£ãŸã®ã§ã¾ã¨ã‚ã¦ã¿ã¾ã—ãŸã€‚æ•£ã€…è¨€ã‚ã‚Œå°½ãã•ã‚Œã¦ã„ã‚‹è©±ã ã¨ã¯æ€ã„ã¾ã™ã€‚ å‰æ è€ƒæ…®ã«å…¥ã‚‹æŽ¡ç”¨åŸºæº– äºˆæ¸¬ç²¾åº¦ (ã‚³ãƒ¼ãƒ‰ã®)ãƒ¡ãƒ³ãƒ†ãƒŠãƒ³ã‚¹ã®å®¹æ˜“æ€§ è¨ˆç®—ã‚ªãƒ¼ãƒ€ãƒ¼ å¦ç¿’æ™‚ äºˆæ¸¬æ™‚ æŒ™å‹•ã®ã‚³ãƒ³ãƒˆãƒãƒ¼ãƒ«ã®ã—ã‚„ã™ã•/äºˆæ¸¬èª¬æ˜Žæ€§ã®å®¹æ˜“ã• ãƒãƒ¥ãƒ¼ãƒ‹ãƒ³ã‚°ã®å¿…è¦æ€§ ãã®ä»– ã¾ã¨ã‚ å‰æ æ©Ÿæ¢°å¦ç¿’ãŒãƒ—ãƒãƒ€ã‚¯ãƒˆã®ä¸»è¦ãªæ¦å™¨ã«ãªã‚‹(ä¾‹ãˆã°æœ€è¿‘è©±é¡Œã«ãªã£ã¦ã„ã‚‹Googleç¿»è¨³ã«ãŠã‘ã‚‹NMT)ã‚‚ã®ã§ã¯ãªãã€ã‚µãƒ¼ãƒ“ã‚¹ã«ãƒ‡ãƒ¼ã‚¿ãŒã¾ãšã‚ã‚Šã€æ©Ÿæ¢°å¦ç¿’ã§ãƒ‡ãƒ¼ã‚¿ã‚’æ´»ç”¨ã™ã‚‹ã“ã¨ã«ã‚ˆã‚Šã€ãã®ã‚µãƒ¼ãƒ“ã‚¹ã‚’æ”¯ãˆã¦ã„ãã‚ˆã†ãªã‚‚ã®ã‚’å‰æã«è€ƒãˆã¦ã„ã¾ã™(ä¾‹ãˆã°CGMã‚µãƒ¼ãƒ“ã‚¹ã®ã‚¹ãƒ‘ãƒ åˆ¤å®š)ã€‚ã¾ãŸã€æŠ•ç¨¿å†…å®¹ã¯ç§å€‹äººã®æ„è¦‹ã§ã‚ã‚Šã€æ‰€å±žçµ„ç¹”ã‚’ä»£è¡¨ã™ã‚‹ã‚‚ã®ã§ã¯ã‚ã‚Šã¾ã›ã‚“ã¨ãŠæ–ã‚Šã—ã¦ãŠã
ihok 2016/11/21
art

google

guideline

æ©Ÿæ¢°å¦ç¿’

programming

project

stat

ml
ãƒªãƒ³ã‚¯
æ©Ÿæ¢°å¦ç¿’ã®åŸºç¤ŽçŸ¥è˜ã¨ã—ã¦ã®æ•°å¦ - learning.ikeay.net
ç§ãŒAIï¼ˆäººå·¥çŸ¥èƒ½ï¼‰ã‚„æ©Ÿæ¢°å¦ç¿’ã£ã¦é›£ã—ã„ãƒŠãƒ¼ã¨æ„Ÿã˜ã‚‹ã¨ã“ã‚ã¯ã€æ•°å¦ã®å‰æçŸ¥è˜ãŒã‚ã‚‹ç¨‹åº¦å¿…è¦ãªã¨ã“ã‚ã§ã™ã€‚ Googleã‹ã‚‰TensorflowãŒå‡ºãŸã¨ãã«ã€ç§ã‚‚ã„ã£ã¡ã‚‡ã‚„ã£ã¦ã¿ã‚‹ã‹ãªã‚“ã¦æ€ã£ãŸã®ã§ã™ãŒã€å‚è€ƒã«ã—ãŸè¨˜äº‹ã‚‚ãªã‹ãªã‹é›£ã—ãã¦ã‚ã‚“ã¾ã‚Šç†è§£ã§ããªã‹ã£ãŸã®ã‚’è¦šãˆã¦ã¾ã™ã€‚é€”ä¸ã¾ã§ç†è§£å‡ºæ¥ã¦ãŸã®ã«ã€æ€¥ã«æ•°å¼ãŒå‡ºã¦ãã¦ã€Œãªã‚‹ã»ã©ã‚ã‹ã‚‰ã‚“ï¼ã€ã£ã¦ãªã‚‹ã“ã¨ãŒå¤šã‹ã£ãŸã§ã™ãã€‚ ã€Œã¨ã„ã†ã‹ã‚¨ãƒ³ã‚¸ãƒ‹ã‚¢ãªã®ã«æ•°å¦è‹¦æ‰‹ãªã®ï½—ã€ã¨ãƒ“ãƒƒã‚¯ãƒªã•ã‚Œã‚‹æ–¹ã‚‚ã„ã‚‰ã£ã—ã‚ƒã‚‹ã¨æ€ã„ã¾ã™ãŒã€ã‚¨ãƒ³ã‚¸ãƒ‹ã‚¢ã£ã¤ã£ãŸã£ã¦ã€ä»Šã®å¾¡æ™‚ä¸–ç†ç³»å‡ºèº«ã‚¨ãƒ³ã‚¸ãƒ‹ã‚¢ã°ã‹ã‚Šã˜ã‚ƒãªã„ã‚“ã§ã™ã€‚ã§ã‚‚ã‚¨ãƒ³ã‚¸ãƒ‹ã‚¢å¥³åã‚„ã£ã¦ã‚‹ã¨è‡ªå‹•ã§ãƒªã‚±ã‚¸ãƒ§æ‰±ã„ã•ã‚Œã‚‹ã‹ã‚‰é¢ç™½ã„ã§ã™ãã€‚ å½“é¢ã®ç›®æ¨™ã¨ã—ã¦ã¯ã€AIã®ä¸ã§ã‚‚æ©Ÿæ¢°å¦ç¿’ã‚’å¦ã‚“ã§ã„ããŸã„ã®ã§ï¼ˆDeepLearningã§ãã‚‹ã‚ˆã†ã«ãªã‚ŠãŸã„ï¼ï¼‰ã€ã‚ã‚‹ã¨å¬‰ã—ã„æ•°å¦ã®çŸ¥è˜ã¨ã—ã¦ã¯ä»¥ä¸‹ã§ã™ã€‚ ç·šå½¢ä»£æ•° ç¢ºçŽ‡ãƒ»çµ±è¨ˆ å¾®åˆ†ãƒ»ç©åˆ† AIã®
ihok 2016/05/11
AI

deeplearning

math

æ©Ÿæ¢°å¦ç¿’

æ•°å¦

ãƒ—ãƒã‚°ãƒ©ãƒŸãƒ³ã‚°

ã‚¨ãƒ³ã‚¸ãƒ‹ã‚¢

programming
ãƒªãƒ³ã‚¯
è¨€èªžå‡¦ç†100æœ¬ãƒŽãƒƒã‚¯ 2015
è¨€èªžå‡¦ç†100æœ¬ãƒŽãƒƒã‚¯ã¯ï¼Œå®Ÿè·µçš„ãªèª²é¡Œã«å–ã‚Šçµ„ã¿ãªãŒã‚‰ï¼Œãƒ—ãƒã‚°ãƒ©ãƒŸãƒ³ã‚°ï¼Œãƒ‡ãƒ¼ã‚¿åˆ†æžï¼Œç ”ç©¶ã®ã‚¹ã‚ãƒ«ã‚’æ¥½ã—ãç¿’å¾—ã™ã‚‹ã“ã¨ã‚’ç›®æŒ‡ã—ãŸå•é¡Œé›†ã§ã™ å®Ÿç”¨çš„ã§ãƒ¯ã‚¯ãƒ¯ã‚¯ã™ã‚‹ã‚ˆã†ãªé¡Œæã‚’åŽ³é¸ã—ã¾ã—ãŸ è¨€èªžå‡¦ç†ã«åŠ ãˆã¦ï¼Œçµ±è¨ˆã‚„æ©Ÿæ¢°å¦ç¿’ãªã©ã®å‘¨è¾ºåˆ†é‡Žã«ã‚‚è¦ªã—ã‚ã¾ã™ ç ”ç©¶ã‚„ãƒ‡ãƒ¼ã‚¿åˆ†æžã®é€²ã‚æ–¹ï¼Œä½œæ³•ï¼Œã‚¹ã‚ãƒ«ã‚’ä¿®å¾—ã§ãã¾ã™ å•é¡Œã‚’è§£ãã®ã«å¿…è¦ãªãƒ‡ãƒ¼ã‚¿ãƒ»ã‚³ãƒ¼ãƒ‘ã‚¹ã‚’é…å¸ƒã—ã¦ã„ã¾ã™ è¨€èªžã¯Pythonã‚’æƒ³å®šã—ã¦ã„ã¾ã™ãŒï¼Œä»–ã®è¨€èªžã«ã‚‚å¯¾å¿œã—ã¦ã„ã¾ã™
ihok 2016/04/22
ãƒ—ãƒã‚°ãƒ©ãƒŸãƒ³ã‚°

python

è‡ªç„¶è¨€èªžå‡¦ç†

Programming

æ©Ÿæ¢°å¦ç¿’

ãƒ‡ãƒ¼ã‚¿

ãƒ‡ãƒ¼ã‚¿åˆ†æž

å¦ç¿’

NLP
ãƒªãƒ³ã‚¯
1