HMM:é ããã«ã³ãã¢ãã« é»åæ å ±å·¥å¦ç§ ä¼åº æå¿ ï® ä¸æ¬¡ãã«ã³ãé£é ï® ç¶æ éå S={1,2,â¦n} ï® é·ç§»ç¢ºç(kâï½) akl ï® é ããã«ã³ãã¢ãã«ï¼Hidden MM:HMMï¼ ï® åºåè¨å·éåΣ ï® åºå確ç(ç¶æ ããåºåè¨å·ã¸ã®åå) ek(b) : S â Σ ãã«ã³ãã¢ãã«ã¨ é ããã«ã³ãã¢ãã« ï® HMMâæéãªã¼ãããã³ï¼ç¢ºç ï® å®ç¾© ï® åºåè¨å·éåΣ ï® ç¶æ éå S={1,2,â¦n} ï® é·ç§»ç¢ºç(kâï½) akl ï® åºå確ç ek(b) ï® éå§ç¶æ çµäºç¶æ 0.4 0.6 0.3 0.7 0.5 0.5 1 2 3 A: 0.2 B: 0.8 A: 0.7 B: 0.3 A: 0.1 B: 0.9 é ããã«ã³ãã¢ãã«(HMM) Rain Dry 0.7 0.3 0.2 0.8 ⢠ï¼ã¤ã®ç¶æ : âRainâ 㨠âDryâ. ⢠æ¨ç§»ç¢ºç:
éºä¼çã¢ã«ã´ãªãºã GA é²åçã¢ã«ã´ãªãºã EA é²åçã¢ã«ã´ãªãºã é²åçã¢ã«ã´ãªãºã ï¼evolutionary algorithmï¼EAã¨ç¥è¨ããã)ã¯é²åçè¨ç® ã®ä¸åéã§ããï¼è¤éãªåé¡ï¼è§£ã®æ¢ç´¢ç©ºéãåºå¤§ã§ããåé¡ï¼ NP-hard ãªåé¡ã®æé©è§£ã®è¿ä¼¼è§£ãæ±ããããã® ã¢ã«ã´ãªãºã ã§ãã åä½ç¾¤ãã¼ã¹ã®ã¡ã¿ãã¥ã¼ãªã¹ãã£ãã¯ãªæé©åã¢ã«ã´ãªãºã ã§ãã çæ®ï¼çªç¶å¤ç°ï¼éºä¼åçµã¿æãï¼èªç¶æ·æ±°ï¼é©è çåï¼ãªã©ã¨ããæ¦å¿µãç¨ãããï¼ ãããçç©é²åã«çæ³ãå¾ãæä½ã§ãããã¨ããï¼é²åè«çè¨ç®ãªã©ã¨å¼ã°ãã¦ãã¾ãã æé©ååé¡ã®è§£ã®åè£ç¾¤ãçç©ã®åä½ç¾¤ã®å½¹å²ãæããï¼ ã³ã¹ãé¢æ°ã«ãã£ã¦ã©ã®è§£ãçãæ®ããã決å®ãï¼ çãæ®ã£ãåä½ç¾¤ã®ä¸ã§äº¤é ãç¹°è¿ãã¾ãããã®éç¨ã§ï¼çªç¶å¤ç°ãå°å ¥ãã¾ãã ããã«ãã£ã¦ããè¯ãåä½ãçãæ®ãã¨ããé²åã«ä¼¼ãéç¨ãèµ·ããã¾ãã ãã®æä½ãç¹°ãè¿ããã¨
ã¡ããã»ã»ã»â èãéãã»ã»ã»â ä»åã¯ãNaive Bayesã¨Complement Naive Bayes, Negation Naive Bayesã§ã®å¤ã¯ã©ã¹åé¡ãæ¯è¼ãããæãã¾ãããªãå¤ã¯ã©ã¹åé¡ã¨ã¯ãææ¸ ãã©ã®ã¯ã©ã¹ ã«å±ãã¦ããããæ±ãããã¨ã§ãã ããããã®ææ³ Naive Bayes Naive Bayesã«ããå¤ã¯ã©ã¹åé¡ã§ã¯ããã¤ãºã®å®çãé©ç¨ããäºå¾ç¢ºç ãæ大ã¨ãªãã¯ã©ã¹ ãæ±ãã¾ãããã ã ãç´æ¥æ¨å®ããã®ã¯é£ããã®ã§ãææ¸ ãææ¸ã«å«ã¾ããåèªå ã§è¿ä¼¼ãã¾ããã¾ããè¨ç®ãç°¡åã«ããããã«åã¯ã©ã¹ã§åèªãç¬ç«ã«çèµ·ããã¨ä»®å®ãã¾ãã ããã¦ãæçµçã«ä»¥ä¸ã®å¼ã«ãã£ã¦ææ¸ã®å±ããã¯ã©ã¹ãæ±ãã¾ãã \begin{align} \tilde{c} = argmax_c P(c) \prod_{i=1}^{n} P(w_i|c) \end{align} Comp
2016/6/25 "第54å ãã¼ã¿ãã¤ãã³ã°+WEB ï¼ æ±äº¬( #TokyoWebmining 54th ) ã¼æ·±å±¤å¦ç¿ã»æ©æ¢°å¦ç¿ ç¥ãã¼" ãéå¬ãã¾ããã 第54å ãã¼ã¿ãã¤ãã³ã°+WEB ï¼ æ±äº¬( #TokyoWebmining 54th ) ã¼æ·±å±¤å¦ç¿ã»æ©æ¢°å¦ç¿ ç¥ãã¼Eventbrite Google ã°ã«ã¼ã ä¼å ´æä¾ãã¦ä¸ãã£ãFreakOutãããã©ãããããã¨ããããã¾ãããç´ æµãªãã¼ã¯ãæä¾ãã¦ãããè¬å¸«ã¡ã³ãã¼ã«æè¬ãã¾ããå¤ãã®æ¹ã ã®åå ãå¬ããæã£ã¦ãã¾ãã åå è IDã»ããã¯ã°ã©ã¦ã³ãä¸è¦§ åå è ã»ãã³ã³ï¼ç¬¬54å ãã¼ã¿ãã¤ãã³ã°+WEB ï¼ æ±äº¬ ã»ãã³ã³ (ä½æãã¦ããã [Twitter:@komiya_atsushi] ããã«æè¬ï¼ 以ä¸ãå ¨è¬å¸«è³æãé¢é£è³æããã¤ã¼ãã¾ã¨ãã§ãã AGENDAï¼ â Opening Talkï¼ O1.ããã¼ã¿ãã¤ã
1 2 3 4 5 6 7 2 X Y X Y 8 9 X X X X i X i 10 11 12 13 Y = f(X) X Y f f 14 15 16 17 18 19 20 21 22 23 Yes Yes No No 24 25 k 26 27 28 29 30 31 32 33 34 KDD Process: Knowledge Discovery and Data Mining 35 36 37 38 39 40 41 42 43 44 45 46 47 48
æ©æ¢°å¦ç¿ã大æµè¡ã ãããæè¿ããæ©æ¢°å¦ç¿ãæ´»ç¨ããÃÃÃãã§ããã¨ããæ©æ¢°å¦ç¿æè¡ãå¿ç¨ããÃÃÃãã¨è¨ã£ãçºè¡¨ããããããããã¨ãããããããã®çºè¡¨ä¼ã«åå ãã¦ã¿ãã¨ãããã£ã¦æ¬å½ã«æ©æ¢°å¦ç¿æè¡ãªã®ï¼ãã¨é ã®ä¸ã«ã¯ã¨ã¹ãã§ã³ãã¼ã¯ãæµ®ãã¶ãããªãã®ãã¡ãã»ããæ©æ¢°å¦ç¿ã¨é«åº¦ãªçµ±è¨å¦çã®éãã¯ãã£ããã©ãã«ããã®ããã¯ãã¾ãæ©æ¢°å¦ç¿ã¯äººå·¥ç¥è½ï¼AIï¼ã®ä¸ç¨®ãªã®ãã ãAIã«ã¯è£å®ç¾©ããã£ã¦ããããå®ç¾ã§ãã¦ãã¾ãã°ãã§ã«AIã§ã¯ãªããªããªãã¦è©±ãããã¾ããæ©æ¢°å¦ç¿ã«ã¤ãã¦ããå®ã¯åããããªé¢ãããã¨æãã¾ããã¨è©±ãã®ã¯ãå½é大å¦GLOCOM åææã§ä¸»ä»»ç 究å¡ã®ä¸è¥¿å´ææ°ã ãä¸è¥¿æ°ã®å°éåéã¯ããã°ãã¼ã¿ããã¼ã¿åæãç¹ã«ç¸é¢åæã«é¢ããæè¡ã®éçºã ãããã«ã¯ã¡ãã£ã¢è«ã人éã®ææ§ãã³ã¢ã¨ãã¦ç°ç¨®ãç°åéã®ååã«ãã価å¤åµçã«é¢ããåé¡ãªã©ã対象ã«ç 究ãè¡ã£ã¦ããã ä¸è¥¿æ°ã«ããã°ãæ©æ¢°å¦
ãªãªã¼ã¹ãé害æ å ±ãªã©ã®ãµã¼ãã¹ã®ãç¥ãã
ææ°ã®äººæ°ã¨ã³ããªã¼ã®é ä¿¡
å¦çãå®è¡ä¸ã§ã
j次ã®ããã¯ãã¼ã¯
kåã®ããã¯ãã¼ã¯
lãã¨ã§èªã
eã³ã¡ã³ãä¸è¦§ãéã
oãã¼ã¸ãéã
{{#tags}}- {{label}}
{{/tags}}