2011/12/27 Nelle Varoquaux <[email protected]>:
> VDHMMs are HMMs in which a state can emit a serie of observations. It seems
> that the main application to these chains are speech recognition, and
> handwriting recognition: fields in which we are trying to determine a state
> to poorly segmented data. Would anyone know other applications than those
> two ? I'm looking for easy test data, in order to illustrate the advantages
> of VDHMMs compared to HMMs.

I hadn't heard of VDHMMs before, but from the description they seem
rather much like the generative cousins of semi-Markov CRFs [1]. Those
are advocated for e.g. named-entity recognition, which is really a
segmentation task, rather than a labeling one (names commonly span
multiple tokens).

[1] www.cs.cmu.edu/~wcohen/postscript/semiCRF.pdf

-- 
Lars Buitinck
Scientific programmer, ILPS
University of Amsterdam

------------------------------------------------------------------------------
Write once. Port to many.
Get the SDK and tools to simplify cross-platform app development. Create 
new or port existing apps to sell to consumers worldwide. Explore the 
Intel AppUpSM program developer opportunity. appdeveloper.intel.com/join
http://p.sf.net/sfu/intel-appdev
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to