æ°ã¯ã¦ãæ£å¼ãªãªã¼ã¹è¨å¿µã¨ãããã¨ã§ããããªãªã¼ã¹ããä½é±éãçµã£ã¡ãã£ããã©ã æ°ã¯ã¦ãªããã¯ãã¼ã¯ã§ã¯ããã¯ãã¼ã¯ã¨ã³ããªãã«ãã´ãªã¸ã¨èªåã§åé¡ãã¦ãããããã®ã«ãã´ãªåé¡ã«ä½¿ããã¦ããã¢ã«ã´ãªãºã ã¯Complement Naive Bayesããããä»æ¥ã¯ãã®ã¢ã«ã´ãªãºã ã«ã¤ãã¦ç´¹ä»ãã¦ã¿ãã Complement Naive Bayesã¯2003å¹´ã®ICMLã§J. Rennieããææ¡ããææ³ã§ãããICMLã¨ããã®ã¯ãæ©æ¢°å¦ç¿ã«é¢ããï¼ãã¶ãï¼æé£é¢ã®å¦ä¼ã§ãæ¡æçã¯ããæ°å¹´ã¯30%ãåã£ã¦ããã2003ã¯119/371ã§ã32.1%ã®æ¡æçã ã£ãããã ã Complement Naive Bayesã®ä½ç½®ã¥ã㯠å®è£ ãç°¡å å¦ç¿æéãçã æ§è½ããããããã ã¨ããæãã§ã2003年段éã«ãã£ã¦ãã絶対çãªæ§è½ã§ã¯SVMã«è² ãã¦ãããããããå¦ç¿ãæ©ãã¨ããã®ã¯å®ã¢ããªã±ã¼ã·
{{#tags}}- {{label}}
{{/tags}}