Discriminative Training Methods for Hidden Markov Models: Theory and Experiments with Perceptron Algorithms

DMCA

Discriminative Training Methods for Hidden Markov Models: Theory and Experiments with Perceptron Algorithms (2002)

Cached

Download Links

by Michael Collins

Citations:

637 - 13 self

BibTeX

@INPROCEEDINGS{Collins02discriminativetraining,
    author = {Michael Collins},
    title = {Discriminative Training Methods for Hidden Markov Models: Theory and Experiments with Perceptron Algorithms},
    booktitle = {},
    year = {2002},
    pages = {1--8}
}

OpenURL

Abstract

We describe new algorithms for training tagging models, as an alternative to maximum-entropy models or conditional random fields (CRFs). The algorithms rely on Viterbi decoding of training examples, combined with simple additive updates. We describe theory justifying the algorithms through a modification of the proof of convergence of the perceptron algorithm for classification problems. We give experimental results on part-of-speech tagging and base noun phrase chunking, in both cases showing improvements over results for a maximum-entropy tagger.

Keyphrases

perceptron algorithm discriminative training method hidden markov model base noun phrase chunking part-of-speech tagging classification problem conditional random field maximum-entropy model viterbi decoding new algorithm simple additive update experimental result training example maximum-entropy tagger

DMCA