Regularization and search for minimum error rate training
Published 1 January 2008
Daniel Cer, Daniel Jurafsky, Christopher D. Manning
Citations45
Generate an AI Snapshot to get a quick, structured summary of this paper.
Study Snapshot
ObjectiveStudy objective
MethodsResearch methodology
PopulationPopulation studied
Sample sizeSample sizes
OutcomesStudy outcomes here
ResultsStudy results comes here
LimitationsResearch study limitations comes here
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
TL;DR
This work contrasts three search strategies for MERT: Powell's method, the variant of coordinate descent found in the Moses MERT utility, and a novel stochastic method.
Abstract
Minimum error rate training (MERT) is a widely used learning procedure for statistical machine translation models. We contrast three search strategies for MERT: Powell's method, the variant of coordinate descent found in the Moses MERT utility, and a novel stochastic method. It is shown that the stochastic method obtains test set gains of +0.98 BLEU on MT03 and +0.61 BLEU on MT05. We also present a method for regularizing the MERT objective that achieves statistically significant gains when combined with both Powell's method and coordinate descent.
Keywords
Computer Science
Moses
4,868 Citations2007Philipp Koehn, Richard Zens +12 more
An open-source toolkit for statistical machine translation whose novel contributions are support for linguistically motivated factors, confusion network decoding, and efficient data formats for translation models and language models.
Computational LinguisticsA Systematic Comparison of Various Statistical Alignment Models
3,928 Citations2003Franz Josef Och, Hermann Ney
An important result is that refined alignment models with a first-order dependence and a fertility model yield significantly better results than simple heuristic models.
Cambridge University Press eBooksNumerical Recipes 3rd Edition: The Art of Scientific Computing
3,439 Citations2007William H. Press, Saul A. Teukolsky +2 more
This new edition incorporates more than 400 Numerical Recipes routines, many of them new or upgraded, and adopts an object-oriented style particularly suited to scientific applications.
Statistical phrase-based translation
3,268 Citations2003Philipp Koehn, Franz Josef Och +1 more
The empirical results suggest that the highest levels of performance can be obtained through relatively simple means: heuristic learning of phrase translations from word-based alignments and lexical weighting of phrase translation.
Minimum error rate training in statistical machine translation
2,766 Citations2003Franz Josef Och
It is shown that significantly better results can often be obtained if the final evaluation criterion is taken directly into account as part of the training procedure.
Discriminative training and maximum entropy models for statistical machine translation
1,044 Citations2001Franz Josef Och, Hermann Ney
A framework for statistical machine translation of natural languages based on direct maximum entropy models, which contains the widely used source-channel approach as a special case and shows that a baseline statistical machinetranslation system is significantly improved using this approach.
A Conditional Random Field Word Segmenter for Sighan Bakeoff 2005
337 Citations2005Huihsin Tseng, Pi-Chuan Chang +3 more
A Chinese word segmentation system built using a conditional random field sequence model that provides a framework to use a large number of linguistic features such as character identity, morphological and character reduplication features is presented.
A unigram orientation model for statistical machine translation
289 Citations2004Christoph Tillmann
The orientation model is shown to improve translation performance over two models: 1) no block re-ordering is used, and 2) the block swapping is controlled only by a language model.
On Some Pitfalls in Automatic Evaluation and Significance Testing for MT
195 Citations2005Stefan Riezler, John T. Maxwell
In an experimental comparison of two statistical significance tests, it is shown that p-values are estimated more conservatively by approximate randomization than by bootstrap tests, thus increasing the likelihood of type-I error for the latter.
Minimum risk annealing for training log-linear models
142 Citations2006David A. Smith, Jason Eisner
This work proposes training log-linear combinations of models for dependency parsing and for machine translation, and describes techniques for optimizing nonlinear functions such as precision or the BLEU metric.
RWTH Publications (RWTH Aachen)A Systematic Comparison of Training Criteria for Statistical Machine Translation
31 Citations2007Richard Zens, Saša Hasan +1 more
It is shown that, not only from a theoretical point of view but also in terms of translation quality, the minimum Bayes risk decision rule is preferable.