Log-Linear Weight Optimization Using Discriminative Ridge Regression Method in Statistical Machine Translation
| Autor: | Francisco Casacuberta, Germán Sanchis-Trilles, Mara Chinea-Rios |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
Log-linear model
Statistical machine translation Machine translation Computer science media_common.quotation_subject 02 engineering and technology computer.software_genre Translation (geometry) Discriminative model Simple (abstract algebra) 0202 electrical engineering electronic engineering information engineering Quality (business) media_common business.industry 020206 networking & telecommunications Pattern recognition Ridge (differential geometry) Regression 020201 artificial intelligence & image processing Discriminative ridge regression Artificial intelligence business LENGUAJES Y SISTEMAS INFORMATICOS computer |
| Zdroj: | Pattern Recognition and Image Analysis ISBN: 9783319588377 IbPRIA RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia instname |
| DOI: | 10.1007/978-3-319-58838-4_4 |
| Popis: | [EN] We present a simple and reliable method for estimating the log-linear weights of a state-of-the-art machine translation system, which takes advantage of the method known as discriminative ridge regression (DRR). Since inappropriate weight estimations lead to a wide variability of translation quality results, reaching a reliable estimate for such weights is critical for machine translation research. For this reason, a variety of methods have been proposed to reach reasonable estimates. In this paper, we present an algorithmic description and empirical results proving that DRR, as applied in a pseudo-batch scenario, is able to provide comparable translation quality when compared to state-of-the-art estimation methods (i.e., MERT [1] and MIRA [2]). Moreover, the empirical results reported are coherent across different corpora and language pairs. The research leading to these results has received funding fromthe Generalitat Valenciana under grant PROMETEOII/2014/030 and the FPI (2014) grant by Universitat Politècnica de València. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|