Parallel FDA5 for Fast Deployment of Accurate Statistical Machine Translation Systems

By Ergun Biçici, Qun Liu, Andy Way. 2014

17 citations recorded by the source corpus.

We use parallel FDA5, an efficiently parameterized and optimized parallel implementation of feature decay algorithms for fast deployment of accurate statistical machine translation systems, taking only about half a day for each translation direction. We build Parallel FDA5 Moses SMT systems for all language pairs in the WMT14 translation task and obtain SMT performance close to the top Moses systems with an average of 3.49 BLEU points difference using significantly less resources for training and development.

Subfields: Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Natural Language Processing

1. Philipp Koehn
2. Lucia Specia
3. Ondrej Bojar
4. Ergun Biçici
5. Barry Haddow
6. Matt Post
7. Qun Liu
8. Rico Sennrich

Citations in the corpus

• Findings of the 2014 Workshop on Statistical Machine Translation (2014, background): “While the precise metric has varied over the years, it has always shared a common idea of computing the average number of times each system was judged better than other systems, and ranking from highest to lowest.”
• Referential Translation Machines for Predicting Translation Quality (2014, methodology): “We use the Parallel FDA5 (Feature Decay Algorithms) instance selection model for selecting the interpretants (Biçici et al., 2014; Biçici and Yuret, 2014) this year, which allows efficient parameterization, optimization, and implementation of FDA, and build an MTPP model (Section 2.1).”
• Referential Translation Machines for Predicting Translation Quality and Related Statistics (2015, methodology): “Instance selection for the training set and the language model (LM) corpus is handled by ParFDA (Bic¸ici et al., 2015), whose parameters are optimized for each translation task.”
• Referential translation machines for predicting semantic similarity (2015, methodology): “More details about the optimization processes are provided in Bic¸ici and Way (2014b) and Bic¸ici et al. (2014).”