AIST4010/ESTR4140: Foundation of Applied Deep Learning — Fall 2026

Course description

This course covers how to use deep learning techniques to resolve real-life computational problems involving different kinds of data. We begin with the deep learning problem-solving paradigm: data preparation, model construction, model training, model evaluation, and hyperparameter search. We then study each part of this paradigm in detail.

The course moves from linear regression toward more complex models. To handle structured data, images, text, sequences, signals, and graphs, we cover CNN/ResNet, RNN/LSTM, attention, and GNN models. We also discuss commonly used techniques for handling overfitting and briefly introduce generative models, including VAE and GAN.

Teaching team

Lecturer: Professor LI Yu (liyu@cse.cuhk.edu.hk), SHB-106. Office hours: Friday, 3:00pm–5:00pm.

TA:

• Pengan Chen (chenpengan@link.cuhk.edu.hk), SHB-904. Office hours: Monday, 4:00pm–6:00pm.
• Jiayi Li (jiayill@link.cuhk.edu.hk), SHB-904. Office hours: Thursday, 10:00am–12:00pm.
• Yuhan Chen (yuhanchen@link.cuhk.edu.hk), SHB-904. Office hours: Wednesday, 2:00pm–4:00pm.

Time and location

Wednesday lecture: 11:30am–1:15pm, Science Centre L4.
Thursday lecture: 4:30pm–5:15pm, T.C. Cheng Bldg C3.
Thursday tutorial: 5:30pm–6:15pm, T.C. Cheng Bldg C3.
ESTR4140 additional Thursday lecture: 3:30pm–4:15pm, T.C. Cheng Bldg C3.

ESTR4140 follows the shared AIST4010 lectures and tutorial above, plus an additional Thursday lecture from 3:30pm to 4:15pm.

Class suspension: No AIST4010 or ESTR4140 lecture or tutorial will be held on Thursday, November 12, due to the CUHK 97th Congregation.

Format

Mainly onsite. Slides will normally be available before each lecture. Online-session information will be posted here when available. Zoom information: Coming soon.

Logistics

Communications

Blackboard is the primary platform for course management and grading. Piazza will be used for class discussion, including anonymous questions. For personal matters, use a private post to the instructor and teaching assistants or contact the teaching team by email.

Piazza details: AIST4010 Piazza.

Grading

• Assignments (42%): Environment setup, written assignments, and Kaggle assignments.
• Scribing (5%): Summarize one lecture and submit it within one week.
• In-class quiz (5%): A short participation-focused quiz; the date will appear in the schedule.
• Project (48%): Proposal (6%), mid-term report (8%), final report (17%), and presentation (17%).

Bonus (up to 2%): one additional scribing assignment (1%) and course surveys (up to 1%). Register through the Scribing preference sheet. Survey statistics are available here.

Open-book quiz policy

The quiz is open book. Detailed arrangements will be announced before the quiz.

Assignment policy

Approximately half of the assignment work uses fixed-answer questions and half uses Kaggle competitions. The final Kaggle competition is optional; students who participate receive the highest two scores among the three Kaggle competitions.

The teaching team will provide linear-regression and simple deep-learning baselines. Performance above these baselines maps to the corresponding score bands, with scores interpolated by ranking among the class.

Release dates and deadlines: See the course schedule.

Programming

All programming assignments must be completed in Python. We recommend using Google Colab.

Scribing

Each student should summarize at least one lecture and submit the note within one week of that lecture. Up to five students may sign up for each lecture. Students may take one additional lecture for a 1% bonus. Notes may be posted online; students may choose whether to display their name.

Registration details: Scribing preference sheet.

Projects

Projects may be completed individually or in a team. Team projects carry higher expectations and should target publication-quality work. The contribution of each member and the workload split must be defined clearly at the beginning. Discuss serious team projects with Professor Li.

The project consists of proposal (6%), mid-term report (8%), final report (17%), and presentation (17%). The instructor contributes 90% of the project assessment and students contribute 10%.

Late days

Each student will have 6 late days to turn in assignments, which can be used on written assignments including A1-written, A2-written, A3-written, project proposal, and project M-report. They cannot be used on Kaggle assignments, the project final report and the scribing note. A maximum of 2 late days can be used for each assignment. Grades will be deducted by 25% for each additional late day. If you would like to use late days for any assignment, please fill in this form before the assignment deadline: Late Day Application.

Post-lecture survey

Deadline for each survey: 11:59pm on the day before the next lecture. We do this because I could have time to answer the questions you mentioned in the survey. Please fill 1 in the Google sheet: Survey results, once you have finished one survey. Usually, we will trust the 1s you fill in the Google sheet. But we will check the things in detail if the number of survey forms we received and the number of 1s on the Google sheet is not consistent.

AI tools

AI tools may be used to polish project reports. Students must submit both their own version and the AI-polished version, and clearly describe how AI tools were used and which parts were affected. A submission without the student's own version is incomplete.

Course schedule and materials

Materials and notes will be added as they are confirmed. All assignment and project deadlines are at 11:59pm unless otherwise stated. The quiz and project presentations take place during the scheduled class.

Tutorial schedule and materials

Tutorial topics are listed below. Materials will be released through this page and Blackboard. Tutorial repository: Coming soon.

Assignments

• Assignment 1: Image
• Assignment 2: Sequence
• Assignment 3: Graph