Tilde’s Machine Translation Systems for WMT 2017

By Marcis Pinnis, Rihards Krišlauks, Toms Miks, Daiga Deksne, Valters Sics. 2018

26 citations recorded by the source corpus.

The paper describes the development process of the Tilde’s NMT systems that were submitted for the WMT 2018 shared task on news translation. We describe the data filtering and pre-processing workflows, the NMT system training architectures, and automatic evaluation results. For the WMT 2018 shared task, we submitted seven systems (both constrained and unconstrained) for English-Estonian and Estonian-English translation directions. The submitted systems were trained using Transformer models.

Subfields: Artificial Intelligence

How researchers see this paper

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

In Artificial Intelligence

Citations in the corpus

Beyond English-Centric Multilingual Machine Translation (2020, neutral): “En ↔ Lv: we report results from Pinnis et al. (2017) on WMT2017.”
Evaluating the morphological competence of Machine Translation Systems (2017, neutral): “In this setup, we measure the degree of certainty to which a system predicts morphological features across small lexical variations. for Latvian and therefore used the LU MII Tagger (Paikens et al., 2013) to parse all Latvian monolingual data available at WMT’17.”
Deep Encoder, Shallow Decoder: Reevaluating the Speed-Quality Tradeoff in Machine Translation (2020, background): “…its ﬁrst proposal (Vaswani et al., 2017), much prior work has assumed that the transformer architecture in machine translation has the same numbers of encoder and decoder layers, including top-performing systems in recent WMT competitions (Edunov et al., 2018; Pinnis et al., 2018; Ng et al., 2019).”
Debugging Neural Machine Translations (2018, background): “For instance, borrowing ideas from parallel corpora ﬁltering [11] such as 1) source-hypothesis sentence length diﬀerence; 2) language identiﬁcation for the hypothesis; 3) digit mismatch between the source and hypothesis; 4) foreign or corrupt symbol checking for the hypothesis.”
On the Linguistic Representational Power of Neural Machine Translation Models (2019, methodology): “2017), but sequence-to-sequence NMT models without explicit syntax are the state-of-the-art at the moment (Sennrich et al. 2017; Pinnis et al. 2017).”