Monday, April 18, 2011

Condtional Random Field - CRF (Machine Learning)

http://en.wikipedia.org/wiki/Conditional_random_field

A conditional random field (CRF) is a type of discriminative undirected probabilistic graphical model. It is most often used for labeling or parsing of sequential data, such as natural language text or biological sequences[1] and computer vision[2] . Specifically, CRFs find applications in shallow parsing[3] , named entity recognition[4] and gene finding, among other tasks, being an alternative to the related hidden Markov models.

http://www.inference.phy.cam.ac.uk/hmw26/crf/

The primary advantage of CRFs over hidden Markov
models is their conditional nature, resulting in the relaxation of the indepen-
dence assumptions required by HMMs in order to ensure tractable inference.

CRFs outperform both MEMMs
and HMMs on a number of real-world sequence labeling tasks


Nando de Freitas
http://www.cs.ubc.ca/~nando/
http://www-stat.stanford.edu/~tibs/ElemStatLearn/download.html


David JC MacKay
http://www.inference.phy.cam.ac.uk/mackay/itila/ 

No comments: