2
votes
HMM and CRF: the label bias problem and I-equivalence
Expanding on my comment: the answer is clearer when you realize that you can't ignore the observed variables. They affect each model differently. As it turns out, the MEMM is not I-equivalent to the ...
- 9,001
Only top scored, non community-wiki answers of a minimum length are eligible