Official name of this course is NPFL095 Modern Methods in Computational Linguistics. It is a continuation of informal Reading Group (RG) meetings. Requirements for getting credits:
- presenting one paper,
- Select a term (write your name to the schedule below) before October 13.
- If no paper is assigned to the term, suggest me 2–3 papers you would like to present (with pdf links, and your preferences) before October 20. Ideally, make a group of 2–4 students presenting papers on a common topic (starting from basics to more advance papers).
- Prepare your presentation and 3–5 quiz questions. At least 3 of the questions should ask for a specific answer, e.g. “write an equation for…”, “given training set X=([dog,N],[cat,Y]), what is the number…” (Not “what do you think about…”). The first question should be quite easy to answer for those who have read the whole paper. The last question may be a tricky one. Send me the questions two weeks before your presentation. We may discuss the paper and refine the questions.
- One week before the presentation, write the questions to a dedicated wiki page here. Send a reminder (questions and a link to the pdf of the paper) to email@example.com by Monday 15:45.
- active participation in the discussions, which is conditioned by reading the papers in advance and attending the meetings,
- sending your answers to me and the presenter by Saturday 23:59 (so the presenter can go through all answers before the presentation and focus more on problematic parts).
- In case of more than three missed meetings or deadlines, additional work (e.g. reports or answers to tricky questions) will be required.
All questions, reports and presented papers must be in English. The presentations are in English by default, but if all present people agree it may be in Czech.
|Oct 4||startup meeting|
|Oct 11||Martin Popel||Kevin Knight, Beáta Megyesi and Christiane Schaefer: The Copiale Cipher, BUCC 2011, you can watch a trailer|
|Oct 18||Daniela Bodanská||Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu: BLEU: a Method for Automatic Evaluation of Machine Translation, ACL 2002. Questions|
|Oct 25||Michael Collins: Discriminative Training Methods for Hidden Markov Models: Theory and Experiments with Perceptron Algorithms, EMNLP 2002. Questions|
|Nov 1||Andrew McCallum, Dayne Freitag, Fernando Pereira: Maximum Entropy Markov Models for Information Extraction and Segmentation, Conference on Machine Learning 2000, slides Question|
| ||—||no RG (Dean's day)|
|Nov 15||John Lafferty, Andrew McCallum, Fernando Pereira: Conditional Random Fields: Probabilistic Models for Segmenting and Labeling Sequence Data, 2001. Questions|
|Nov 22||Tomas Mikolov, Kai Chen, Greg Corrado, Jeffrey Dean: Efficient Estimation of Word Representations in Vector Space, ICLR 2013|