Materials

• slides
• Our lecture, seminar (russian)
• CMU lectures (english): intro, seq-labeling

More materials

• Tutorial on Conditional Random Fields - url
• Neural sequence tagging with CRF loss - blog post
• More CMU lectures on applied structured learning: now with CRF

Practice

Your task for this week is to implement POS statistical tagger model. You can choose one of the several options

• If you prefer notebooks, pick tagger.ipynb. Submit the notebook with your results
• If you want good old-fashioned python, pick tagger.py instead. Submit that file along with your training results (text or plots)

There's also an optional task, rnn_tagger.ipynb - it attempts the same task with neural networks. You are not required to do that task. However, if you got through it, you are eligible for some bonus points AND a a great opportunity to beat classical methods with deep learning :)