The seminar will focus on building a strong machine translation system using neural based encoder-decoder networks completely from scratch. The speaker will go over the full pipeline starting from the details of well-established practices on preprocessing the data that is fed into model and ending on decoding from trained model and evaluating it compared to ground truth. The audience’s familiarity with PyTorch https://pytorch.org
will be helpful as the presentation will include some code pieces written in it. At the end the speaker will give his high-level strategy on how to adopt the current systems to English <-> Azerbaijani machine translation.
Elman Mansimov is Ph.D. student at New York University. His research concerns machine learning, in particular neural networks and their various applications. His undergraduate degree is in Computer Science from University of Toronto. He has done internships in Google Brain, Facebook AI Research and Apple.
Date: Jan 11, 2019
Venue: ADA University, Room B203
Open to the public.