40959 Deep Generative Models

Course Description

Deep Generative Models

Course Information

Required Texts

  1. [BSH] Bishop, Christopher M. and Hugh Bishop, Deep Learning: Foundations and Concepts, Springer, 2024.

  2. [MUR] Murphy, Kevin P, Probabilistic Machine Learning: Advanced Topics, The MIT Press, 2023.

  3. [TOM] Tomczak, Jakub M., Deep Generative Modeling, Springer, 2024.

Grading Policy

  1. 20%: Mid-term exam (1404/01/25).

  2. 20%: Final exam (1404/03/26).

  3. 35%: Homeworks.

  4. 15%: Quiz.

  5. 10%: Paper & Explore a theoretical or empirical question and present it. Deadline for choosing paper: 1404/01/25.

  6. 5%: Class Activity

Lecture Schedule


Lecture Lecture Date Topics Related Readings and Links Homeworks & Assignments Quizes
1 1403-11-20Introduction: what is deep generative models? Chapter 20 of MUR
Chapter 1 of TOM
2
3
1403-11-27
1403-11-29
Structured density Chapter 10 of MUR
Chapter 11 of BSH
4 1403-12-04Disentangled Representation Learning Papers given in the slides
5 1403-12-06Autoregressive Generative Models Chapter 22 of MUR
Chapter 2 of TOM
6
7
8
1403-12-11
1403-12-13
1403-12-18
Varational Autoencoder Chapter 21 of MUR
Chapter 4 of TOM
Papers given in the slides


Quiz
9
10
11
12
1403-12-20
1403-12-25
1404-01-16
1404-01-18
Generative Adverserial networksChapter 26 of MUR
Chapter 7 of TOM
Papers given in the slides

Quiz

13 1404-01-23 Flow-based models Chapter 22 of MUR
Chapter 3 of TOM
14 1404-01-25 Mid-term exam Presentation topic selection deadline
15 1404-01-30 Flow-based models Papers given in the slides
16
17
1404-02-01
1404-02-06
Energy-based models Chapter 24 of MUR
Chapter 3 of TOM
Papers given in the slides
18
19
1404-02-08
1404-02-13
Score-based models Papers given in the slides
20
21
22
1404-02-15
1404-02-17
1404-02-22
Diffusion Models Chapter 25 of MUR
Chapter 9 of TOM
Papers given in the slides