The NLU+ course consists of three live lectures a week which are also live streamed and recorded for those who are unable to attend in person. To view lectures go to Learn Ultra and click on the Lecture Recordings link.
In all cases the lecture content and the required reading are examinable, the background reading is not.
Slides | Lecture | Course Content |
1 (All) | Required reading: The Future of Computational Linguistics: Beyond the Alchemy Church and Lieberman (21) | |
2 (Birch) | Background reading: Automating Knowledge Acquisition for Machine Translation, Knight. | |
| 3 (Birch) | Required reading: Word Alignment and the Expectation Maximization Algorithm, Lopez. Alternative to required reading: A tutorial MT workbook, Knight. Gives a different view of similar material. Contains exercises, so helpful if you want to revise some basic ideas about probability. Background reading:
|
There are now many introductions to the basics of neural networks for NLP, which we'll focus on during the first few weeks of the course. What matters is that you understand the underlying concepts, rather than a particular author's notation. Everyone learns differently, so if you prefer a different take on perceptrons, feedforward models, recurrent networks, or attention, you may want to consult some of these resources in addition to (or as alternative to) the required readings:
- Speech and Language Processing (3rd edition draft), Jurafsky & Martin.
- Natural Language Processing, Eisenstein.
- Deep Learning, Goodfellow et al.