User Tools

Site Tools


keynote:2011-lesson04

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
keynote:2011-lesson04 [2011/08/23 11:44]
11021025
keynote:2011-lesson04 [2014/05/22 08:34] (current)
Line 64: Line 64:
 ** 4.4 最大似然估计 ** ** 4.4 最大似然估计 **
  
-从一个给定的O和Q中,似然值为:\\ +  ​从一个给定的O和Q中,似然值为:\\ 
-$L(A,​B,​\pi)=a_{i_1}b_{i_1o_1}a_{i_1i_2}b_{i_2o_}...a_{i_{T-1}i_T}b_{i_To_T}$\\ +  $L(A,​B,​\pi)=a_{i_1}b_{i_1o_1}a_{i_1i_2}b_{i_2o_}...a_{i_{T-1}i_T}b_{i_To_T}$\\ 
-Log-likehood 值为\\ +  Log-likehood 值为\\ 
-$l(A,​B,​\pi)=\sum_{i=1}^Mf_{i0}ln(a_i)+\sum_{i=1}^M\sum_{j=1}^Mf_{ij}ln(a_{ij})+\sum_{i=1}^M\sum_{o(i)}ln(b_{io})$\\ +  $l(A,​B,​\pi)=\sum_{i=1}^Mf_{i0}ln(a_i)+\sum_{i=1}^M\sum_{j=1}^Mf_{ij}ln(a_{ij})+\sum_{i=1}^M\sum_{o(i)}ln(b_{io})$\\ 
-最大似然估计就是要求以下参量:\\ +  最大似然估计就是要求以下参量:\\ 
-$a_i=\frac{f_{i0}}{1} a_ij=\frac{f_{ij}}{\sum_{j=1}^Mf_{ij}}$\\+  $a_i=\frac{f_{i0}}{1} a_ij=\frac{f_{ij}}{\sum_{j=1}^Mf_{ij}}$\\
    
-由于直接从似然函数求最大似然估计过于困难,人们采用一些技术来计算: +  ​由于直接从似然函数求最大似然估计过于困难,人们采用一些技术来计算: 
-- The Segmental K-means Algorith +    - The Segmental K-means Algorith 
-- The Baum-Welch (E-M) Algorithm+   ​- The Baum-Welch (E-M) Algorithm
  
  
keynote/2011-lesson04.txt · Last modified: 2014/05/22 08:34 (external edit)