Dušan Variš

Main Research Interests

 

• Neural Machine Translation
• Sequential Task Learning in Neural Networks
• Modularity in Neural Networks
• Japanese-English Translation

 

Projects

Current Projects:

• MASAPI - Multilingual Assistant for Searching, Analysing and Processing Information and decision support
• HPLT - High Performance Language Technologies
• LINDAT Translation - LINDAT translation service - model training and production deployment
• Formal Linguistics Seminar - recording of lectures (archive)

 

​Previous Projects:

• Bergamot - Browser-based Multilingual Translation - multi-source machine translation
• SSHOC - Social Sciences & Humanities Open Cloud - machine translation and text processing support
• ELG - European Language Grid - dockerization of translation models
• Neural Monkey - An open-source tool for sequence learning in NLP built on TensorFlow.
• Health In My Language - machine translation of medical texts from English to Czech, German, Romanian and Polish
• MTViet - Machine Translation from Vietnamese into Czech for the Purposes of the Police of the Czech Republic
• MLFix - automatic machine translation post-editing tool based on statistical machine learning (successor of Depfix)
• Treex in Docker - dockerization of the up-to-date revision of the Treex NLP framework

Curriculum Vitae

My CV (in English) is available here.

Teaching

(past) NSWI095 Introduction to Unix (labs), more info.

Selected Bibliography

• Google Scholar
• ORCID: 0000-0003-4408-2186
• Scopus ID: 56582304500
• Researcher ID: P-4531-2017

Selected Talks:

• Are Transformers Good Learners? Exploring the Limits of Transformer Training (video)

Bibliography:

• Google Scholar
• dbpl
• Biblio.