The University of Maryland statistical machine translation system for the Fourth Workshop on Machine Translation

Dyer, Chris; Setiawan, Hendra; Marton, Yuval; Resnik, Philip

doi:10.3115/1626431.1626461

Cited by 5 publications

(15 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In previous shared tasks of WMT, there have been submissions that use other metrics for tuning (e.g. Dyer et al 2009) in order to achieve higher correlation with human judgment. In our experiments, however, tuning on Meteor or Meteor-SCP can be better than tuning on Bleu even if our aim is to obtain a higher Bleu score (cf.…”

Section: Results and Analysismentioning

confidence: 99%

See 1 more Smart Citation

Metric and reference factors in minimum error rate training

Way

2010

Machine Translation

View full text Add to dashboard Cite

In Minimum Error Rate Training (MERT), Bleu is often used as the error function, despite the fact that it has been shown to have a lower correlation with human judgment than other metrics such as Meteor and Ter. In this paper, we present empirical results in which parameters tuned on Bleu may lead to sub-optimal Bleu scores under certain data conditions. Such scores can be improved significantly by tuning on an entirely different metric altogether, e.g. Meteor, by 0.0082 Bleu or 3.38% relative improvement on the WMT08 English-French data. We analyze the influence of the number of references and choice of metrics on the result of MERT and experiment on different data sets. We show the problems of tuning on a metric that is not designed for the single reference scenario and point out some possible solutions.

show abstract

Section: Results and Analysismentioning

confidence: 99%

“…He, A. Way also been some efforts to tune against other criteria, such as Bleu−Ter 2 (Dyer et al 2009) or the IQ MT metric (Lambert et al 2006).…”

Section: Introductionmentioning

confidence: 99%

Metric and reference factors in minimum error rate training

Way

2010

Machine Translation

View full text Add to dashboard Cite

show abstract

“…Although sparse features are useful, training of sparse features is an extremely difficult optimization problem, and at this point there is still no method that has been widely demonstrated as being able to robustly estimate the parameters of millions of features. Because of this, a third approach of first training the parameters of sparse features, then condensing the sparse features into dense features and performing one more optimization pass (potentially with a different algorithm), has been widely used in a large number of research papers and systems (Dyer et al 2009;He and Deng 2012;Flanigan, Dyer, and Carbonell 2013;Setiawan and Zhou 2013). A dense feature created from a large group of sparse features and their weights is generally called a summary feature, and can be expressed as follows h sum ( f , e, d) = w sparse h sparse ( f , e, d)…”

Section: Summary Featuresmentioning

confidence: 99%

“…Finally, there is also some work on optimizing multiple evaluation metrics at one time. The easiest way to do so is to simply use the linear interpolation of two or more metrics as the error function (Dyer et al 2009; He and Way 2009; Servan and Schwenk 2011):…”

Section: Evaluation Measures and Optimizationmentioning

confidence: 99%

Optimization for Statistical Machine Translation: A Survey

Neubig

Watanabe

2016

Computational Linguistics

View full text Add to dashboard Cite

In statistical machine translation (SMT), the optimization of the system parameters to maximize translation accuracy is now a fundamental part of virtually all modern systems. In this article, we survey 12 years of research on optimization for SMT, from the seminal work on discriminative models (Och and Ney 2002) and minimum error rate training (Och 2003), to the most recent advances. Starting with a brief introduction to the fundamentals of SMT systems, we follow by covering a wide variety of optimization algorithms for use in both batch and online optimization. Specifically, we discuss losses based on direct error minimization, maximum likelihood, maximum margin, risk minimization, ranking, and more, along with the appropriate methods for minimizing these losses. We also cover recent topics, including large-scale optimization, nonlinear models, domain-dependent optimization, and the effect of MT evaluation measures or search on optimization. Finally, we discuss the current state of affairs in MT optimization, and point out some unresolved problems that will likely be the target of further research in optimization for MT.

show abstract

“…We begin by adapting the lattice technique of Dyer et al (2009) to Finnish. We train a standard phrasebased machine translation model on a new corpus: on the source side we concatenate the original data with its one-best segmentation according to a Morfessor (Creutz and Lagus, 2007) model trained on the original data, and on the target side we simply concatenate it with itself.…”

Section: Finnish Tokenization Using Morfessor and Word-latticesmentioning

confidence: 99%

The University of Illinois submission to the WMT 2015 Shared Translation Task

Schwartz

Bryce²,

Geigle

et al. 2015

Proceedings of the Tenth Workshop on Statistical Machine Translation

View full text Add to dashboard Cite

In this year's WMT translation task, Finnish-English was introduced as a language pair of competition for the first time. We present experiments examining several variations on a morphologically-aware statistical phrase-based machine translation system for translating Finnish into English. Our system variations attempt to mitigate the issue of rich agglutinative morphology when translating from Finnish into English. Our WMT submission for Finnish-English preprocesses Finnish data with omorfi (Pirinen, 2015), a Finnish morphological analyzer. We also present results for two other language pairs with morphologically interesting source languages, namely German-English and Czech-English.

show abstract

The University of Maryland statistical machine translation system for the Fourth Workshop on Machine Translation

Cited by 5 publications

References 20 publications

Metric and reference factors in minimum error rate training

Metric and reference factors in minimum error rate training

Optimization for Statistical Machine Translation: A Survey

The University of Illinois submission to the WMT 2015 Shared Translation Task

Contact Info

Product

Resources

About