- Sep 6: Welcome all the students! Wish you enjoy our NLP course and have a splendid semester!
- Wednesday, 4:30-6:15pm, Y.C. Liang Hall 101 Thursday, 4:30-5:15pm, Hui Yeung Shing Bldg G01
- Wednesday, 1:30-2:15pm, William M W Mong Eng Bldg 407
- Prof. Liwei WANG
- Teaching Assistant
- Yanyang LI Mark ZHAO
- All updates will be sent to your emails. Please check your university emails.
- Assignments + Paper Presentation + Final Project
Tutorial sessions would be the fastest and clearest way to discuss any problem. You are also welcomed to send emails to the instructor and the TAs.
Natural language processing (NLP) is a crucial part of artificial intelligence (AI), which aims to endow computers with the ability to process and understand human language. This course gives an overview of modern deep learning techniques for natural language processing. The course starts with basic linguistic concepts in NLP and moves from shallow bag-of-words representations to richer structural embeddings, which is the foundation for the successful use of deep learning in NLP. Then the course will guide you through three fundamental tasks of NLP: language modeling (LM), natural language understanding (NLU), and natural language generation (NLG), followed by some recent advances. Along the way we will introduce cutting-edge computational models together with insights from a linguistic perspective.
All related lecture materials will be posted on Blackboard → Lecture Notes.
The following schedule (tentative) are subject to minor modifications.
|1||Sep 8, 9||Intro & Basic Knowledge about NLP|
|2||Sep 15, 16||Neural Networks|
|3||Sep 23||Word and Word Vectors: part 1||The day following the Chinese Mid-Autumn Festival
Release of Assignment 1
|4||Sep 29, 30||Word and Word Vectors: part 2;
|Submission of Assignment 1
Complete team forming for projects
Assign selected paper to each team
|5||Oct 6, 7||Transformers & Pretraining|
|6||Oct 13||More About Transformers & Pretrained Models
(Team Presentation on Selected Papers)
|Double Ninth Festival|
|7||Oct 20, 21||Question Answering||Release of Assignment 2
Release of Final Project
|8||Oct 27, 28||Machine Translation||Submission of Assignment 2|
|9||Nov 3, 4||NLG Tasks||Release of Assignment 3|
|10||Nov 10, 11||Language + Vision||Submission of Assignment 3|
|11||Nov 17, 18||Integrating Knowledge into Large Models|
|12||Nov 24, 25||Frontiers Topics in NLP|
|13||Dec 1, 2||Final Projects||Submission of Final Project|
There will be three assignments and one paper presentation in total.
- Three Assignments (Question Sets or Coding)
- Paper Presentation
All three assignments must be done independently by yourself.
If there is no coding involved, please submit your answers as a single PDF file. Otherwise, please compress the PDF file with related codes into a ZIP file. The PDF file should be accepted by VeriGuide before submission. All the submissions should be made via Blackboard by 11:59pm
You are encouraged to either typeset your solution in LaTeX or Word, or scan or take photos of your handwritten solution (please bundle your scans/photos into a single PDF file)
Paper presentation will be carried out by teams of four.
Each team will be randomly assigned one state-of-the-art paper from the paper pool at Week 4. Read, study, and discuss the paper thoroughly and present the paper to the class during lecture time at Week 6.
To register a team, please send an email from the team captain to TAs and cc all the team members before week 4.
More information about Final Project will be released later.
To maximize the project achievement, students should form teams of four in general. We recommend you using the same team as in paper presentation.
You are encouraged include in your report any illustrative, concrete examples that help you understand the papers you read.
Please submit your project report (in PDF) together with related codes and material online as a single zip file at blackboard.
- Academic Honesty Policy
- Late Submission Policy
The Chinese University of Hong Kong places very high importance on honesty in academic work submitted by students, and adopts a policy of zero tolerance on academic dishonesty.
All submissions of assignments, project, etc. should be processed by VeriGuide first.
For all the assignments and projects, late submission are subject to score deduction:
- Late by less than 12 hours -- deducted by 20%
- Late by 12 to 24 hours -- deducted by 40%
- Late by 24 to 36 hours -- deducted by 60%
- Late by 36 to 48 hours -- deducted by 80%
- Submissions that are late by more than 48 hours will not be accepted.