Re: forward backward algorithm - HMM
- From: Ted Dunning <ted.dunning@xxxxxxxxx>
- Date: Wed, 29 Aug 2007 01:25:30 GMT
You'll need to give it the observations, the state transition
probabilities, and the observation probabilities. This tutorial may
help if you haven't already seen it:http://www.cs.cmu.edu/~roni/11761-s07/Presentations/whw-HMMs-in-ASR-3...
Jason is a bit confused.
You need to give the DECODING algorithm all of the data that he
specifies, but the TRAINING algorithm only gets the observations.
(That is the whole point of the training)
[ comp.ai is moderated ... your article may take a while to appear. ]
.
- References:
- forward backward algorithm - HMM
- From: Alessia
- Re: forward backward algorithm - HMM
- From: Jason Adams
- forward backward algorithm - HMM
- Prev by Date: Hybrid Systems
- Next by Date: CFP: 2nd Workshop on Statistical and Machine learning approaches applied to ARchitectures and compilaTion (SMART'08), Goteborg, Sweden
- Previous by thread: Re: forward backward algorithm - HMM
- Next by thread: CFP: GPEM: Special Issue on Evolutionary Computation in Games
- Index(es):