Universität Wien
Achtung! Das Lehrangebot ist noch nicht vollständig und wird bis Semesterbeginn laufend ergänzt.

340199 VU Advanced Machine Translation (2023W)

5.00 ECTS (3.00 SWS), SPL 34 - Translationswissenschaft
Prüfungsimmanente Lehrveranstaltung

An/Abmeldung

Hinweis: Ihr Anmeldezeitpunkt innerhalb der Frist hat keine Auswirkungen auf die Platzvergabe (kein "first come, first served").

Details

max. 25 Teilnehmer*innen
Sprache: Englisch

Lehrende

Termine (iCal) - nächster Termin ist mit N markiert

Mittwoch 11.10. 16:45 - 20:00 Medienlabor II ZfT Gymnasiumstraße 50 4.OG
Mittwoch 25.10. 16:45 - 20:00 Medienlabor II ZfT Gymnasiumstraße 50 4.OG
Mittwoch 08.11. 16:45 - 20:00 Medienlabor II ZfT Gymnasiumstraße 50 4.OG
Mittwoch 22.11. 16:45 - 20:00 Medienlabor II ZfT Gymnasiumstraße 50 4.OG
Mittwoch 06.12. 16:45 - 20:00 Medienlabor II ZfT Gymnasiumstraße 50 4.OG
Mittwoch 10.01. 16:45 - 20:00 Medienlabor II ZfT Gymnasiumstraße 50 4.OG
Mittwoch 24.01. 16:45 - 20:00 Medienlabor II ZfT Gymnasiumstraße 50 4.OG

Information

Ziele, Inhalte und Methode der Lehrveranstaltung

- This course will be taught in person in the ZTW computer lab, with a hybrid or online option.

Goals:
Students will acquire specialised and practical knowledge on neural machine translation (NMT), word alignment models, self-attention architectures, multilingual NMT, domain adaptation approaches for NMT, and NMT decoding.
Using state-of-the-art technologies, students will learn to implement unsupervised word alignment methods, and apply different approaches to customise multilingual NMT models.

Content:

- Multilingual NMT
- Domain adaptation for multilingual NMT
- Efficient transformer architectures
- Decoding for NMT
- Unsupervised word alignment models

Didactic approach:
Students will need to complete practical assignments involving a range of approaches for word alignment, multilingual NMT, and domain adaptation. Students will also gain experience of unsupervised, deep generative models, and decoding for NMT. The course will be taught in English, with some opportunities for using other languages to complete the coursework.

Art der Leistungskontrolle und erlaubte Hilfsmittel

Continuous evaluation:
- Weekly reflections, and paper presentations count for 40% of the mark.
- Word alignment task deliverable counts for 20% of the mark.
- Benchmarking of domain adaptation approaches for NMT task deliverable counts for 40% of the mark.

Mindestanforderungen und Beurteilungsmaßstab

In order to pass this module, a student needs to reach the threshold of 4.
MT marking map
excellent - sehr gut (1)
good - gut (2)
average - befriedigend (3)
sufficient - genügend (4)
insufficient - nicht genügend (5)

Prüfungsstoff

- MT word alignment
- Self-attention architectures
- Multilingual NMT
- Multilingual NMT domain adaptation
- NMT decoding

Literatur

Core texts:
- Kenny, Dorothy. 2022. Machine translation for everyone: Empowering users in the age of artificial intelligence. (Translation and Multilingual Natural Language Processing 18). Berlin: Language Science Press. DOI: 10.5281/zenodo.6653406 (url: https://langsci-press.org/catalog/book/342)
- Koehn, P. 2020. Neural Machine Translation. Cambridge University Press
- Peter F. Brown, Stephen A. Della Pietra, Vincent J. Della Pietra, and Robert L. Mercer. 1993. The Mathematics of Statistical Machine Translation: Parameter Estimation. Computational Linguistics, 19(2):263–311.
- Collins, Michael. “Statistical Machine Translation : IBM Models 1 and 2.” (2011).
- Yi Tay, Mostafa Dehghani, Dara Bahri, and Donald Metzler. 2022. Efficient Transformers: A Survey. ACM Comput. Surv. 55, 6, Article 109 (June 2023), 28 pages. https://doi.org/10.1145/3530811
- Danielle Saunders. 2022. Domain Adaptation for Neural Machine Translation. In Proceedings of the 23rd Annual Conference of the European Association for Machine Translation, pages 9–10, Ghent, Belgium. European Association for Machine Translation.

Additional recommended resources:
- Visualizing A Neural Machine Translation Model (Mechanics of Seq2seq Models With Attention). https://jalammar.github.io/visualizing-neural-machine-translation-mechanics-of-seq2seq-models-with-attention/
- Voita, Lena. Neural Machine Translation Inside Out. https://lena-voita.github.io/posts/nmt_inside_out.html
- Lopez, Adam. Word Alignment and the Expectation-Maximization Algorithm. https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=8338914856defebe908394c2b33dc43d350c5dd0
- Huang, A., Subramanian, S., Sum, J., Almubarak, K., Biderman, S., & Rush, S. (2022). The Annotated Transformer.(2022). URL https://nlp.seas.harvard.edu/annotated-transformer/

Zuordnung im Vorlesungsverzeichnis

Letzte Änderung: Mi 27.09.2023 10:48