7. Hidden Markov Models (Parte 2) (pdf, it, 413 KB, 4/28/10)
7. Hidden Markov Models (Parte 2) (pdf, it, 413 KB, 4/28/10)
7. Hidden Markov Models (Parte 2) (pdf, it, 413 KB, 4/28/10)
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Model selection: solutions<br />
• Typical solution (usable for many probabilistic<br />
models)<br />
– train several models w<strong>it</strong>h different orders k<br />
– choose the one maximizing an “optimal<strong>it</strong>y” cr<strong>it</strong>erion<br />
Which “optimal<strong>it</strong>y” cr<strong>it</strong>erion?<br />
• First naive solution: maximizing likelihood of<br />
data w.r.t. model<br />
Maximizing Log Likelihood<br />
• Problem: Log Likelihood is not decreasing<br />
when augmenting the order<br />
-50<br />
-<strong>10</strong>0<br />
-150<br />
-200<br />
-250<br />
0 5 <strong>10</strong> 15 20<br />
Not applicable cr<strong>it</strong>erion!<br />
54<br />
55<br />
4