EE5438 Semester B 2024/2025

Applied Deep Learning

Master deep learning to lead in the AI era.

Course Description

In this hands-on course, Applied Deep Learning, you will delve into the practical applications of deep neural networks, exploring both the theoretical foundations and real-world implementations of cutting-edge AI models. Through a combination of lectures and Python-based group projects utilizing frameworks like PyTorch, you will gain the essential skills to design, train, fine-tune, and deploy deep learning architectures to solve complex problems in AI. By the end of the course, you will be equipped to apply state-of-the-art deep learning techniques to build innovative solutions and tackle real-world challenges.

Topics:

  • A Brief History of AI with Deep Learning
  • Perceptron and Multi-Layer Perceptrons (MLPs)
  • Gradient Descent and Backpropagation
  • Optimizations and Regularizations
  • Convolutional Neural Networks (CNNs)
  • Reurrent Neural Networks (RNNs)
  • Self-Attention and Transformers
  • Large Language Models (LLMs): BERT, GPT and T5
  • LLM Decoding Strategies and Prompt Engineering
  • Parameter-Efficient Fine-Tuning (PEFT)
  • Instruction Tuning and Preference Alignment
  • Multimodality: CLIP, Flamingo, LLaVA, GPT-4V

Grading:

  • Assignments - 15%
  • Quiz - 5%
  • Midterm - 10%
  • Group Project - 20%
  • Final Exam - 50%

Prerequisites:

  • Multi-variable Calculus and Linear Algebra
  • Probability and Statistics
  • Object-Oriented Programming

Please be advised, EE5438 is a course with STRONG Mathematical and Programming components.

Instructor:

Prof. PO Lai-Man

Graders:

WU Haoxuan
LIU Yuyang
LI Kun

Level and Units:

  • P5 Level with 3 CUs


Lecture:

  • Monday : 12:00pm to 2:50pm

Venue:

  • Run Run Shaw Creative Media Centre
  • CMC M5050

Rules:

  • Lecture attendance: Required. You are responsible for whatever is taught in the lecture and tutorial.

  • Submission of Assignments and Project Reports:
    1. Hardcopies are not required.
    2. Soft copies of assignment and project materials, such as PDF files, presentation PPTs, Colab source code and related files, need to be uploaded to CANVAS by the deadline of 11:00pm.

  • Late Submission:
    1. Assignments and project reports mush be received by the course email account before 11:00PM on the due date.
    2. NO late homework is accepted without previous arrangement with the instructor.
    3. If approved, late homework receives 20% per business day penalty.
    4. Write your name and student number on the top right corner of the paper.
    5. Write your answers legibly and derive all of the steps to show your work clearly.
    6. No credits will be given to answers without showing the steps.
    7. Students may work together on the homework, but copying is unacceptable.

  • Cheating: In particular copying your fellow classmate's assignments or programs, is a very serious offense! If you are found cheating, you will automatically get an F grade in this course and your act will be reported to the Department for necessary disciplinary actions. Please don't let others copy your assignments or programs as we don't have a way to tell who is copying who and you may be liable to the penalties.