activity
20162024
most citedHow Language-Neutral is Multilingual BERT?

76 citations · 144 across the 23 of their papers we have counts for

collaborators
Showing 2020 · cs.CLShow all

7 papers · 2 filters

cs.CL2020

The LMU Munich System for the WMT 2020 Unsupervised Machine Translation Shared Task

Alexandra Chronopoulou, Dario Stojanovski, Viktor Hangya +1

This paper describes the submission of LMU Munich to the WMT 2020 unsupervised shared task, in two language directions, German<->Upper Sorbian. Our core unsupervised neural machine…

cs.CL2020★ 1 cited

Reusing a Pretrained Language Model on Languages with Limited Corpora for Unsupervised NMT

Alexandra Chronopoulou, Dario Stojanovski, Alexander Fraser

Using a language model (LM) pretrained on two languages with large monolingual data in order to initialize an unsupervised neural machine translation (UNMT) system yields state-of-…

cs.CL2020

Anchor-based Bilingual Word Embeddings for Low-Resource Languages

Tobias Eder, Viktor Hangya, Alexander Fraser

Good quality monolingual word embeddings (MWEs) can be built for languages which have large amounts of unlabeled text. MWEs can be aligned to bilingual spaces using only a few thou…

cs.CL2020

Pragmatic information in translation: a corpus-based study of tense and mood in English and German

Anita Ramm, Ekaterina Lapshinova-Koltunski, Alexander Fraser

Grammatical tense and mood are important linguistic phenomena to consider in natural language processing (NLP) research. We consider the correspondence between English and German t…

cs.CL2020

Addressing Zero-Resource Domains Using Document-Level Context in Neural Machine Translation

Dario Stojanovski, Alexander Fraser

Achieving satisfying performance in machine translation on domains for which there is no training data is challenging. Traditional supervised domain adaptation is not suitable for…

cs.CL2020

Towards Reasonably-Sized Character-Level Transformer NMT by Finetuning Subword Systems

Jindřich Libovický, Alexander Fraser

Applying the Transformer architecture on the character level usually requires very deep architectures that are difficult and slow to train. These problems can be partially overcome…