Deep Generative Models
Tuesday 16.15 - 18.00 Delta (Narva mnt 18) - 1022
This course is based on materials from Stanford CS236 Deep Generative Models course developed by Stefano Ermon and Aditya Grover and available at deepgenerativemodels.github.io under MIT License.
CS236 is a deep dive into generative models. It will be useful for those students who not only want to try out GANs, but to understand probabilistic foundations and learning algorithms behind them. The course discusses applications of generative models for natural language processing, image recognition, graph mining, and reinforcement learning.
Topics covered in the course:
- Autoregressive Models
- Variational Autoencoders
- Generative Adversarial Networks
- Normalizing Flow Models
Prerequisites
We expect all of the course participants to know the topics covered in MTAT.03.227 Machine Learning. You should be comfortable with:
- linear algebra,
- calculus,
- probability theory,
- Python (on a good level)
This is going to be a tough course, so do not take the prerequisites lightly.
Organization
We will use the reverse classroom approach, where we read the material at home and discuss it in the class. Discussions will be led by students themselves.
Each participant (possibly in groups of two, depending on the number of attendees) will:
- deliver a presentation based on the CS236 lecture slides (~45 min)
- solve and explain to the class an exercise from the homework (~30 min)
- make a quiz based on the covered material (~10 min, 5-10 questions)
Timings are approximate, for the presentation we encourage you to prepare your own slides. Note! Currently lecture recordings are not available, but they might appear in future. In this case we will slightly change the course model to watch videos at home and discuss them in class.
There will be a lot of self-organization in this course, be ready!
Grading
To pass the course students need to prepare one seminar (lecture + exercise), prepare the quiz and score at least 60% of points from all other quizzes.
Contacts
Mikhail Papkov (mikhail.papkov at ut.ee) - instructor
Prof. Raul Vicente (raul.vicente.zafra at ut.ee) - coordinating the course