[B! ãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚º] seikennã®ãƒ

seikenn id:seikenn

ãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚ºã«é–¢ã™ã‚‹seikennã®ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ (9)

${{author_name}}$

{{{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™ã‚¤ã‚ºåˆ†é¡žã‚’ãƒ™ãƒ¼ã‚¹ã«ã—ãŸSmartNewsã®ãƒãƒ£ãƒ³ãƒãƒ«åˆ¤å®š | SmartNewsé–‹ç™ºè€…ãƒ–ãƒã‚°
æ ªå¼ä¼šç¤¾ã‚´ã‚¯ãƒã®ä¸è·¯ã§ã™ã€‚æ™®æ®µã¯æ©Ÿæ¢°å¦ç¿’ã®æ‰‹æ³•ã‚’ç”¨ã„ãŸã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ æ”¹å–„ãªã©ã€ã‚µãƒ¼ãƒãƒ¼ã‚µã‚¤ãƒ‰ã®é–‹ç™ºã‚’è¡Œã£ã¦ã„ã¾ã™ã€‚ SmartNewsã§ã¯æ§˜ã€…ãªãƒ‹ãƒ¥ãƒ¼ã‚¹è¨˜äº‹ã‚’ã€Œã‚¨ãƒ³ã‚¿ãƒ¡ã€ã€Œã‚¹ãƒãƒ¼ãƒ„ã€ã€Œã‚°ãƒ«ãƒ¡ã€ãªã©ã®ãƒãƒ£ãƒ³ãƒãƒ«ã«åˆ†ã‘ã¦è¡¨ç¤ºã—ã¦ã„ã¾ã™ã€‚ãã®ã‚ˆã†ãªã“ã¨ã‚’å¯èƒ½ã«ã™ã‚‹ãŸã‚ã«ã¯ã€ãƒ‹ãƒ¥ãƒ¼ã‚¹è¨˜äº‹ãŒã©ã®ãƒãƒ£ãƒ³ãƒãƒ«ã«å±žã™ã‚‹ã®ã‹ã‚’åˆ¤æ–ã™ã‚‹å¿…è¦ãŒã‚ã‚‹ã‚ã‘ã§ã™ãŒã€ãã‚Œã‚’è¡Œã£ã¦ã„ã‚‹ã®ã¯äººã§ã¯ã‚ã‚Šã¾ã›ã‚“ã€‚æ©Ÿæ¢°ãŒã€ã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ ã«åŸºã¥ã„ã¦ã€è‡ªå‹•çš„ã«è¡Œã£ã¦ã„ã¾ã™ã€‚ ä»Šå›žã®ã‚¨ãƒ³ãƒˆãƒªãƒ¼ã§ã¯ã€ãã®ã€Œè‡ªå‹•çš„ã«ãƒãƒ£ãƒ³ãƒãƒ«ã«åˆ†é¡žã™ã‚‹ä»•çµ„ã¿ã€ã«ã¤ã„ã¦æ›¸ã“ã†ã¨æ€ã„ã¾ã™ã€‚ SmartNewsã«ãŠã‘ã‚‹ã€ãƒ‹ãƒ¥ãƒ¼ã‚¹è¨˜äº‹ã®ãƒãƒ£ãƒ³ãƒãƒ«åˆ¤å®šã‚’å˜ç´”åŒ–ã™ã‚‹ã¨ã€ãƒ™ãƒ¼ã‚¹ã«ã¯ã€ŒãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚ºåˆ†é¡žå™¨ã€ã¨å‘¼ã°ã‚Œã‚‹ã€æ©Ÿæ¢°å¦ç¿’ã®åˆæ©çš„ãªçŸ¥è¦‹ãŒã‚ã‚Šã¾ã™ã€‚ã“ã®ã‚¨ãƒ³ãƒˆãƒªãƒ¼ã§ã¯ãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚ºåˆ†é¡žå™¨ã‚’ãƒ¡ã‚¤ãƒ³ãƒˆãƒ”ãƒƒã‚¯ã¨ã—ã¦å–ã‚Šä¸Šã’ã¾ã™ã€‚ãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚ºåˆ†é¡žå™¨ã«ã¤ã„ã¦ã¯ã€ã™ã§ã«æ§˜ã€…ãªã¨ã“
seikenn 2013/07/24
æ©Ÿæ¢°å¦ç¿’

ãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚º
ãƒªãƒ³ã‚¯
Naive Bayes Text Classifier
Text classification by Naive Bayes alpha= Positive Negative åˆ†é¡žçµæžœ å¦ç¿’çµæžœ å˜èªžå°¤åº¦ (word likelihood) å˜èªžé »åº¦ (word frequency)
seikenn 2011/09/27
ãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚º
ãƒªãƒ³ã‚¯
ä»Šã•ã‚‰ãªã„ãƒ¼ã¶ã¹ã„ãš - kisa12012ã®æ—¥è¨˜
ä»Šé€±ã®è¨€èªžå‡¦ç†ã®ãŸã‚ã®æ©Ÿæ¢°å¦ç¿’å…¥é–€ã‚’èªã‚€ä¼šï¼ˆç ”ç©¶å®¤å†…å‹‰å¼·ä¼šï¼‰ã¯ï¼ŒãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚ºã®ç« ã«çªå…¥ï¼Ž ã©ã†ã›ãªã‚‰å¾©ç¿’ã¤ã„ã§ã«C++ã§å®Ÿè£…ã—ã‚ˆã†ã‹ãªï¼Œã¨ãªã‚“ã¨ãªãæ€ã„ç«‹ã£ãŸã®ã§å®Ÿè£…ã—ã¦ã¿ã¾ã—ãŸï¼Ž ã‚³ãƒ¼ãƒ‰ GitHub - oiwah/classifier: Online Classification Library ã€Œè¨€èªžå‡¦ç†ã®ãŸã‚ã®ã€æ©Ÿæ¢°å¦ç¿’ï¼Œãªã®ã§ã‚¯ãƒ©ã‚¹ã‚„ç‰¹å¾´ã¯ã™ã¹ã¦stringã§å…¥åŠ›ã•ã‚Œã‚‹ã“ã¨ã‚’æƒ³å®šã—ãŸå®Ÿè£…ã«ã—ã¦ã„ã¾ã™ï¼Žã¨ã„ã†ã‹ï¼Œæ±Žç”¨åŒ–ã™ã‚‹ã®ãŒé¢d(ry çµæžœ training data Positive good bad good good Positive good excellent good excellent Positive exciting good excellent exciting boring Negative bad good bad Negative excellent
seikenn 2011/05/21
ãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚º

æ©Ÿæ¢°å¦ç¿’
ãƒªãƒ³ã‚¯
Complement NaiveBayesã‚’å®Ÿè£…ã—ãŸã‚ˆ - kisa12012ã®æ—¥è¨˜
ãƒ¬ãƒƒãƒ‰ãƒ–ãƒ«ã¨ã‚«ãƒ¬ãƒ¼ãŒç¾Žå‘³ã—ã„å£ç¯€ã«ãªã‚Šã¾ã—ãŸï¼Ž å‰å›žï¼ŒãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚ºã‚’å®Ÿè£…ã—ãŸå¾Œï¼Œ ã€Œã©ã†ã›ãªã‚‰Complement NaiveBayesã‚‚å®Ÿè£…ã—ã¦ã—ã¾ã„ãªã‚ˆï¼Žã€ ã¨ã„ã†å¤©ã‹ã‚‰ã®å£°ãŒèžã“ãˆãŸæ°—ãŒã—ãŸã®ã§ï¼Œå®Ÿè£…ã—ã¦ã¿ã¾ã—ãŸï¼Ž Complement NaiveBayesã¨ã¯ãªã‚“ãžã‚„ï¼Œã¨ã„ã†æ–¹ã¯ï¼Œä»¥ä¸‹ã®è¨˜äº‹ã§éžå¸¸ã«ä¸å¯§ã«è§£èª¬ã•ã‚Œã¦ã„ã‚‹ã®ã§ï¼Œãã¡ã‚‰ã‚’å‚ç…§ãã ã•ã„ï¼Ž ã“ã¡ã‚‰ã§ã‚‚ç°¡å˜ã«èª¬æ˜Žã™ã‚‹ã¨ï¼ŒComplement NaiveBayesã¯ãã®ã‚¯ãƒ©ã‚¹ã«ã€Œå±žã—ãªã„ã€è¨˜äº‹ã‚’ç”¨ã„ã¦ï¼Œæ–‡æ›¸ã«å¯¾ã™ã‚‹å°¤åº¦ã‚’è¨ˆç®—ã—ã¾ã™ï¼Žãã—ã¦ï¼Œå°¤åº¦ãŒä¸€ç•ªã€Œä½Žã„ã€ã‚¯ãƒ©ã‚¹ã‚’äºˆæ¸¬çµæžœã¨ã—ã¦å‡ºã™æ‰‹æ³•ã§ã™ï¼ŽNaiveBayesã¨åå¯¾ã§ã™ãï¼Žãã®æ€§è³ªä¸Šï¼Œ2ã‚¯ãƒ©ã‚¹ã®å ´åˆã¯NaiveBayesã¨Complement NaiveBayesã¯çµæžœãŒä¸€è‡´ã—ã¾ã™ï¼Ž æ–°ã¯ã¦ãªãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ã§ã‚‚ä½¿ã‚ã‚Œã¦ã‚‹Complement Naive Bayesã‚’è§£èª¬ã™ã‚‹ã‚ˆ
seikenn 2011/05/21
ãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚º

æ©Ÿæ¢°å¦ç¿’
ãƒªãƒ³ã‚¯
ãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚ºã¨ã¯ï¼Ÿ - BayesWiki
ç·¨é›†ç”¨ãƒ¦ãƒ¼ã‚¶ãƒ¼å: bayes ãƒ‘ã‚¹ãƒ¯ãƒ¼ãƒ‰:bayes ã‚³ãƒ¡ãƒ³ãƒˆã‚’æ›¸ãã¨ãã«ã¯ã€ãŠåå‰æ¬„ã« bayes ã‚’æ›¸ãåŠ ãˆã¦ãã ã•ã„ã€‚ ã—ã‚ƒã¹ã‚Šå ´6:è°è«–ãƒ»é›‘è«‡ã¯ã“ã¡ã‚‰ QandA:è³ªå•ã¯ã“ã¡ã‚‰ æƒ…å ±æä¾›æ¿ãã®ä¸€ï¼šæƒ…å ±æä¾› BookReview/ å’Œæ›¸, æ´‹æ›¸ ãƒªãƒ³ã‚¯é›†: (æ—§)ãƒ†ãƒ¼ãƒžåˆ¥ä¼šè°å ´ ( ç¢ºçŽ‡ã¨ã¯ä½•ã‹ï¼Ÿ / ãƒ™ã‚¤ã‚ºçš„ã¨ã¯ã©ã†ã„ã†ã“ã¨ã‹ï¼Ÿ / æ©Ÿæ¢°ã¯æ„å‘³ã‚’ç†è§£ã§ãã‚‹ã‹ï¼Ÿ / ç”»åƒã®äº‹å‰åˆ†å¸ƒã¨ã¯ï¼Ÿ / ãƒ™ã‚¤ã‚ºçš„ãƒ¢ãƒ‡ãƒ«ã®å…ˆç«¯ / ç”Ÿç‰©ã¯ãƒ™ã‚¤ã‚ºæŽ¨å®šã—ã¦ã„ã‚‹ã‹ï¼Ÿ ) æœ€æ–°ã®20ä»¶ 2010-08-30 MATLABæŽ²ç¤ºæ¿ 2010-08-24 ã—ã‚ƒã¹ã‚Šå ´6 2009-11-20 æ±Žç”¨æŽ²ç¤ºæ¿2006å¹´6æœˆ 2009-11-04 æƒ…å ±æä¾›æ¿ãã®ä¸€ 2009-10-23 QandA 2009-09-09 ã—ã‚ƒãƒ™ãƒªæš6 2008-11-07 è³ªå•ç®±ï¼ã“ã¨ã°ã®æ„å‘³é™å®š 2008-09-03 Rec
seikenn 2011/03/04
ãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚º
ãƒªãƒ³ã‚¯
FrontPage - BayesWiki
ãŠçŸ¥ã‚‰ã› â€ ãƒ™ã‚¤ã‚ºã‚¦ã‚£ã‚ã®ç½®ãå ´ã‚’NAISTï¼ˆhttp://hawaii.aist-nara.ac.jp/~shige-o/pukiwiki/index.phpï¼‰ ã‹ã‚‰äº¬éƒ½å¤§å¦ã®ã‚µãƒ¼ãƒãƒ¼ä¸Š ï¼ˆhttp://hawaii.sys.i.kyoto-u.ac.jp/~oba/bayeswiki/index.phpï¼‰ã«ç§»å‹•ã—ã¾ã—ãŸ. 2008/08/01 ã‚¹ãƒ‘ãƒ å¯¾ç–ã«è¨ç½®ã—ãŸä½µè¨ä¼šè°å ´ã®ã»ã†ã®ã‚¹ãƒ‘ãƒ ãŒã²ã©ããªã£ã¦ããŸã®ã§ã€ ã‚¹ãƒ‘ãƒ å¯¾ç–ã‚’å¼·åŒ–ã—ãŸæœ¬å®¶pukiwiki ã®ã»ã†ã‚’ãƒ¡ã‚¤ãƒ³ã¨ã™ã‚‹ã¹ãæ•´å‚™ä¸ã§ã™ã€‚ 2007/05/18 ã‚³ãƒ¡ãƒ³ãƒˆã‚¹ãƒ‘ãƒ å¯¾ç–ã®ãŸã‚ã«ã€åˆè¨€è‘‰ã‚’è¨å®šã—ã¾ã—ãŸã€‚ ã‚³ãƒ¡ãƒ³ãƒˆã‚’æ›¸ãè¾¼ã‚€ã¨ãã«ã¯ ãŠåå‰æ¬„ã«è‡ªåˆ†ã®ãƒãƒ³ãƒ‰ãƒ«åã¨ã¨ã‚‚ã« bayes ã¨åŠè§’è‹±å—ã§æ›¸ãè¾¼ã‚€ ã‚ˆã†ã«ã—ã¦ä¸‹ã•ã„ã€‚ 2007/05/09 æŠ•ç¨¿ã‚„ç·¨é›†ã«ãƒ‘ã‚¹ãƒ¯ãƒ¼ãƒ‰ã‚’è¦æ±‚ã™ã‚‹ã‚ˆã†ã«ã—ã¾ã—ãŸã€‚ ãƒ¦ãƒ¼ã‚¶ãƒ¼åã€ãƒ‘ã‚¹ãƒ¯ãƒ¼ãƒ‰ã¨ã‚‚ã«è‹±å°æ–‡
seikenn 2011/02/28
ãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚º
ãƒªãƒ³ã‚¯
Text Classification with CEEK.JP NEWS
CEEK.JP NEWS ã®2009å¹´1æœˆã‹ã‚‰2011å¹´12æœˆã®è¨˜äº‹ãƒ‡ãƒ¼ã‚¿ã‚’åŸºã«ã€ãƒ†ã‚ã‚¹ãƒˆã®åˆ†é¡žã‚’è¡Œã„ã¾ã™ã€‚å¯¾å¿œã—ã¦ã„ã‚‹ã‚«ãƒ†ã‚´ãƒªãƒ¼ã¯ã€Œç¤¾ä¼šã€ã€Œæ”¿æ²»ã€ã€Œå›½éš›ã€ã€ŒçµŒæ¸ˆã€ã€Œé›»è„³ã€ã€Œã‚¹ãƒãƒ¼ãƒ„ã€ã€Œã‚¨ãƒ³ã‚¿ãƒ¼ãƒ†ã‚¤ãƒ¡ãƒ³ãƒˆã€ã€Œã‚µã‚¤ã‚¨ãƒ³ã‚¹ã€ã®8ã¤ã§ã™ã€‚ ãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚ºï¼ˆNaive Bayesï¼‰åˆ†é¡žæ³•ã‚’ç”¨ã„ã¦ãŠã‚Šã€æ¦‚ã80%ã®ç²¾åº¦ã§åˆ†é¡žã§ãã¾ã™ã€‚ ã‚³ãƒ³ãƒ•ãƒ¥ãƒ¼ã‚¸ãƒ§ãƒ³ãƒ»ãƒžãƒˆãƒªãƒƒã‚¯ã‚¹ï¼ˆå¦ç¿’ï¼š2005å¹´7æœˆ / åˆ¤å®šï¼š2005å¹´8æœˆï¼‰ http://labs.ceek.jp/classify/cm.pdf è¡¨ã®ç¸¦ï¼ˆã‚°ãƒ©ãƒ•ï¼‰ã¯ã€æŽ¨å®šåˆ†é‡Žã€‚è¡¨ã®æ¨ªã¯ã€æ£è§£åˆ†é‡Žã€‚ å‚è€ƒè³‡æ–™ï¼š æƒ…å ±æ„å‘³è«–ï¼ˆç¬¬8å›žï¼‰ ãƒ™ã‚¤ã‚ºå¦ç¿’ ï¼ˆæ«»äº•ç ”ç©¶å®¤ æƒ…å ±æ„å‘³è«–ã®è¬›ç¾©è³‡æ–™ï¼‰ Tackling the Poor Assumptions of Naive Bayes Text Classifiers
seikenn 2011/01/17
ãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚º

ãƒ†ã‚ã‚¹ãƒˆåˆ†é¡ž
ãƒªãƒ³ã‚¯
ãƒ™ã‚¤ã‚ºå¦ç¿’ã®ç†è«–
ã¨ã„ã†æŒ™å‹•ã‚’æŒã¡ã¾ã™ã€‚Î»ã‚’å¦ç¿’ä¿‚æ•°ã¨å‘¼ã¶ã“ã¨ã«ã™ã‚‹ã¨ã€å¦ç¿’ãƒ¢ãƒ‡ãƒ«ã¨äº‹å‰åˆ†å¸ƒãŒ ä¸Žãˆã‚‰ã‚ŒãŸã¨ãã€Î»ã‚’æ±‚ã‚ã‚‹ã“ã¨ãŒå¦ç¿’ç†è«–ã®ç¬¬ä¸€æ©ã«ç›¸å½“ã—ã¾ã™ã€‚ â—Žå®Ÿå•é¡Œã§ã¯ q(x) ãŒã‚ã‹ã‚‰ãªã„ã®ã§ã€ç†è«–çš„ã«å¾—ã‚‰ã‚ŒãŸ G ã‚ã‚‹ã„ã¯ G ã¨é–¢ä¿‚ã®ã‚ã‚‹å€¤ã¨ã€å®Ÿé¨“çš„ã« å¾—ã‚‰ã‚ŒãŸå€¤ã¨ã‚’æ¯”è¼ƒã—ã¦ q(x)ãŒä½•ã§ã‚ã‚‹ã‹ã‚’è€ƒãˆã¾ã™ã€‚ã“ã®æ“ä½œã‚’ãƒ¢ãƒ‡ãƒ«ã®é¸æŠžã‚ã‚‹ã„ã¯æ¤œå®šã¨ã„ã„ã¾ã™ã€‚ ãã®ãŸã‚ã®åŸºç¤Žã¨ã—ã¦ G ã‚’çŸ¥ã‚‹å¿…è¦ãŒã‚ã‚Šã¾ã™ã€‚ç¢ºçŽ‡å¤‰æ•° G ã‚ã‚‹ã„ã¯ã€G ã«é–¢é€£ã—ãŸé‡ã®æŒ™å‹•ã‚’è§£æ˜Žã™ã‚‹ã®ãŒå¦ç¿’ç†è«–ã§ã™ã€‚ â—Žã€Œå®Ÿå•é¡Œã§ã¯ q(x) ãŒã‚ã‹ã‚‰ãªã„ã‹ã‚‰ç†è«–ã¯å½¹ç«‹ãŸãªã„ã€ã¨æ„Ÿã˜ã‚‹äººãŒã‚ã‚‹ã‚ˆã†ã§ã™ãŒã€ ãã‚Œã¯åå¯¾ã§ã™ã€‚ã€Œå®Ÿå•é¡Œã§ã¯ q(x) ãŒã‚ã‹ã‚‰ãªã„ã‹ã‚‰ã€ç†è«–çš„ã«å¾—ã‚‰ã‚ŒãŸçµæžœã¨å®Ÿé¨“çµæžœã‚’ æ¯”è¼ƒã—ã¦ q(x) ã«ã¤ã„ã¦è€ƒãˆã‚‹ã“ã¨ãŒã§ãã‚‹ã€ã®ã§ã™ã€‚ã“ã‚Œã¯å¦ç¿’ç†è«–ã«é™ã‚‰ãšã€å…¨ã¦ã®ç§‘å¦ã«ãŠã„ã¦ åŸºæœ¬çš„ãªè€ƒãˆæ–¹ã§ã‚ã‚‹ã¨æ€ã„
seikenn 2010/11/09
ãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚º
ãƒªãƒ³ã‚¯
æ–°ã¯ã¦ãªãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ã§ã‚‚ä½¿ã‚ã‚Œã¦ã‚‹Complement Naive Bayesã‚’è§£èª¬ã™ã‚‹ã‚ˆ - å°„æ’ƒã—ã¤ã¤å‰è»¢ æ”¹
æ–°ã¯ã¦ãƒ–æ£å¼ãƒªãƒªãƒ¼ã‚¹è¨˜å¿µã¨ã„ã†ã“ã¨ã§ã€‚ã‚‚ã†ãƒªãƒªãƒ¼ã‚¹ã‹ã‚‰ä½•é€±é–“ã‚‚çµŒã£ã¡ã‚ƒã£ãŸã‘ã©ã€‚ æ–°ã¯ã¦ãªãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ã§ã¯ãƒ–ãƒƒã‚¯ãƒžãƒ¼ã‚¯ã‚¨ãƒ³ãƒˆãƒªã‚’ã‚«ãƒ†ã‚´ãƒªã¸ã¨è‡ªå‹•ã§åˆ†é¡žã—ã¦ã„ã‚‹ãŒã€ã“ã®ã‚«ãƒ†ã‚´ãƒªåˆ†é¡žã«ä½¿ã‚ã‚Œã¦ã„ã‚‹ã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ ã¯Complement Naive Bayesã‚‰ã—ã„ã€‚ä»Šæ—¥ã¯ã“ã®ã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ ã«ã¤ã„ã¦ç´¹ä»‹ã—ã¦ã¿ã‚‹ã€‚ Complement Naive Bayesã¯2003å¹´ã®ICMLã§J. Rennieã‚‰ãŒææ¡ˆã—ãŸæ‰‹æ³•ã§ã‚ã‚‹ã€‚ICMLã¨ã„ã†ã®ã¯ã€æ©Ÿæ¢°å¦ç¿’ã«é–¢ã™ã‚‹ï¼ˆãŸã¶ã‚“ï¼‰æœ€é›£é–¢ã®å¦ä¼šã§ã€æŽ¡æŠžçŽ‡ã¯ã“ã“æ•°å¹´ã¯30%ã‚’åˆ‡ã£ã¦ã„ã‚‹ã€‚2003ã¯119/371ã§ã€32.1%ã®æŽ¡æŠžçŽ‡ã ã£ãŸã‚ˆã†ã ã€‚ Complement Naive Bayesã®ä½ç½®ã¥ã‘ã¯ å®Ÿè£…ãŒç°¡å˜ å¦ç¿’æ™‚é–“ãŒçŸã„ æ€§èƒ½ã‚‚ãã“ãã“ã‚ˆã„ ã¨ã„ã†æ„Ÿã˜ã§ã€2003å¹´æ®µéšŽã«ã‚ã£ã¦ã‚‚ã€çµ¶å¯¾çš„ãªæ€§èƒ½ã§ã¯SVMã«è² ã‘ã¦ã„ãŸã€‚ã—ã‹ã—ã€å¦ç¿’ãŒæ—©ã„ã¨ã„ã†ã®ã¯å®Ÿã‚¢ãƒ—ãƒªã‚±ãƒ¼ã‚·
seikenn 2010/11/09
ãƒŠã‚¤ãƒ¼ãƒ–ãƒ™ã‚¤ã‚º
ãƒªãƒ³ã‚¯
1