The Machine Translation Group at CIS held two lectures on machine translation. The state-of-the-art until 2016 is phrase-based statistical machine translation, which uses word aligned parallel corpora to learn how to translate blocks of words (called phrases). The second approach is the current state-of-the-art and is related to deep learning. Deep Learning is an interesting new branch of machine learning where neural networks consisting of multiple layers have shown new generalization capabilities. Neural Machine Translation (NMT) is a new paradigm in data-driven machine translation. In Neural Machine Translation, the entire translation process is posed as an end-to-end supervised classification problem, where the training data is pairs of sentences and the full sequence to sequence task is handled in one model.
Email Address: SubstituteLastName@cis.uni-muenchen.de
Email Address: SubstituteFirstInitialSubstituteLastName@cis.uni-muenchen.de
| Date | Paper | Links | Presenter |
| Thursday, January 10th | Phrase-based Machine Translation | slides | Matthias Huck |
| Thursday, January 17th | NMT and Research at CIS on Translation to Morphologically Rich Languages | Morph Rich slides NMT slides | Alexander Fraser |
Further literature:
Philipp Koehn's book Statistical Machine Translation
Luong, Cho, Manning 2016 ACL Tutorial on Neural Machine Translation
Sennrich, Haddow 2017 EACL Tutorial on Neural Machine Translation