Optimization for Statistical Machine Translation: A Survey

Neubig, Graham; Watanabe, Takahiro

doi:10.1162/coli_a_00241

Cited by 16 publications

(13 citation statements)

References 82 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Estimating the parameters of a MT system from the rewards can not be done with the usual MT optimization methods: as the reference is not known, it is impossible to score a n-best list as required by methods optimizing a classification criterion such as MERT or MIRA (Neubig and Watanabe, 2016). Moreover, as only one translation hypothesis is scored, methods optimizing a ranking criterion, such as PRO, can also not be used.…”

Section: Optimizing a Mt System From Weak Feedbackmentioning

confidence: 99%

LIMSI Submission for WMT'17 Shared Task on Bandit Learning

Wisniewski¹

2017

Proceedings of the Second Conference on Machine Translation

View full text Add to dashboard Cite

This paper describes LIMSI participation to the WMT'17 shared task on Bandit Learning. The method we propose to adapt a seed system trained on out-domain data to a new, unknown domain relies on two components. First, we use a linear regression model to exploit the weak and partial feedback the system receives by learning to predict the reward a translation hypothesis will get. This model can then be used to score hypotheses in the search space and translate source sentences while taking into account the specificities of the in-domain data. Second, we use the UCB1 algorithm to choose which of the 'adapted' or 'seed' system must be used to translate a given source sentence in order to maximize the cumulative reward.Results on the development and train sets show that the proposed method does not succeed in improving the seed system. We explore several hypotheses to explain this negative result.

show abstract

Section: Optimizing a Mt System From Weak Feedbackmentioning

confidence: 99%

LIMSI Submission for WMT'17 Shared Task on Bandit Learning

Wisniewski¹

2017

Proceedings of the Second Conference on Machine Translation

View full text Add to dashboard Cite

show abstract

“…Therefore, our work can be classed at the intersection of two research disciplines; The former is the optimization of decoder parameters for machine translation [5,6]. Where, for the majority of decoders, the objective function combines log-linearly a set of translation features to evaluate the translation hypotheses.…”

Section: Related Workmentioning

confidence: 99%

“…In the machine translation community to optimise these weights, the proposed algorithms are largely based on a grid search algorithm [6]. Where the goal is to find the best set of weights which minimise a loss function adapted for the translation process [5].…”

Section: Related Workmentioning

confidence: 99%

A Translation Evaluation Function based on Neural Network

Douib¹,

Langlois²,

Smaïli³

2017

View full text Add to dashboard Cite

Abstract. In this paper, we study the feasibility of using a neural network to learn a fitness function for a machine translation system based on a genetic algorithm termed GAMaT. The neural network is learned on features extracted from pairs of source sentences and their translations. The fitness function is trained in order to estimate the BLEU of a translation as precisely as possible. The estimator has been trained on a corpus of more than 1.3 million data. The performance is very promising: the difference between the real BLEU and the one given by the estimator is equal to 0.12 in terms of Mean Absolute Error.

show abstract

“…Conventional MT systems, be they phrase-based, n-gram based, syntax-based or hierarchical, are typically trained in two steps: the first step (training) estimates individual features functions; the second one (tuning) learns to combine these features so as to optimize translation quality, for instance using Minimum Error Rate Training (MERT) (Och, 2003). The limitations of MERT, notably its inability to train feature sets containing more than a dozen of features, have long been reported, and more effective discriminative training procedures have been sought (see (Neubig and Watanabe, 2016) for a recent review).…”

Section: Related Workmentioning

confidence: 99%

A comparison of discriminative training criteria for continuous space translation models

Allauzen

Khanh

Yvon

2017

Machine Translation

View full text Add to dashboard Cite

This paper explores a new discriminative training procedure for continuousspace translation models (CTMs) which correlates better with translation quality than conventional training methods. The core of the method lays in the definition of a novel objective function which enables us to effectively integrate the CTM with the rest of the translation system through N-best rescoring. Using a fixed architecture, where we iteratively retrain the CTM parameters and the log-linear coefficients, we compare various ways to define and combine training criteria for each of these steps, drawing inspirations both from max-margin and learning-to-rank techniques. We experimentally show that a recently introduced loss function, which combines these two techniques, outperforms several objective functions from the literature. We also show that ensuring the consistency of the losses used to train these two sets of parameters is beneficial to the overall performance.

show abstract

Optimization for Statistical Machine Translation: A Survey

Cited by 16 publications

References 82 publications

LIMSI Submission for WMT'17 Shared Task on Bandit Learning

LIMSI Submission for WMT'17 Shared Task on Bandit Learning

A Translation Evaluation Function based on Neural Network

A comparison of discriminative training criteria for continuous space translation models

Contact Info

Product

Resources

About