Conditional random fields: Probabilistic models for segmenting and labeling sequence data

DMCA

Conditional random fields: Probabilistic models for segmenting and labeling sequence data (2001)

Cached

Download Links

Other Repositories/Bibliography

by John Lafferty

Citations:

3482 - 85 self

BibTeX

@INPROCEEDINGS{Lafferty01conditionalrandom,
    author = {John Lafferty},
    title = {Conditional random fields: Probabilistic models for segmenting and labeling sequence data},
    booktitle = {},
    year = {2001},
    pages = {282--289},
    publisher = {Morgan Kaufmann}
}

OpenURL

Abstract

We present conditional random fields, a framework for building probabilistic models to segment and label sequence data. Conditional random fields offer several advantages over hidden Markov models and stochastic grammars for such tasks, including the ability to relax strong independence assumptions made in those models. Conditional random fields also avoid a fundamental limitation of maximum entropy Markov models (MEMMs) and other discriminative Markov models based on directed graphical models, which can be biased towards states with few successor states. We present iterative parameter estimation algorithms for conditional random fields and compare the performance of the resulting models to HMMs and MEMMs on synthetic and natural-language data. 1.

Keyphrases

conditional random field probabilistic model sequence data label sequence data stochastic grammar towards state directed graphical model maximum entropy markov model natural-language data hidden markov model successor state present iterative parameter estimation algorithm several advantage strong independence assumption fundamental limitation discriminative markov model

DMCA