Difference between revisions of "CH391L/HiddenMarkovModel"

From Marcotte Lab
Jump to: navigation, search
Line 7: Line 7:
  
 
As you see, both emission probability and transition probability are given.  
 
As you see, both emission probability and transition probability are given.  
<pre>[Emission probability]
+
<pre>
 +
[Emission probability]
 
Fair:  1/6 1/6 1/6 1/6 1/6 1/6
 
Fair:  1/6 1/6 1/6 1/6 1/6 1/6
 
Loaded: 0.1 0.1 0.1 0.1 0.1 0.5
 
Loaded: 0.1 0.1 0.1 0.1 0.1 0.5
Line 15: Line 16:
 
F -> L : 0.05
 
F -> L : 0.05
 
L -> L : 0.90
 
L -> L : 0.90
L -> F : 0.10</pre>
+
L -> F : 0.10
 +
</pre>
 
----
 
----
 
[[Category:CH391L]]
 
[[Category:CH391L]]

Revision as of 11:27, 19 February 2011

Viterbi Algorithm

This is an 'dishonest casino' example from Durbin, et al. 'Biological Sequence Analysis' book. A casino used two types of dice: one is a 'fair' one that has equal chance to all six numbers, and the other is a 'loaded' one that has high chance of number 6 than the others. We have a sequence of dice numbers from the casino, and want to estimate when a dealer exchanged the dice.

We assume that the Hidden Markov Model of this example as below:

HMM_dice.png

As you see, both emission probability and transition probability are given.

[Emission probability]
Fair:   1/6 1/6 1/6 1/6 1/6 1/6
Loaded: 0.1 0.1 0.1 0.1 0.1 0.5

[Transition probability]
F -> F : 0.95
F -> L : 0.05
L -> L : 0.90
L -> F : 0.10