Exercise 7 - Generative Modelling

Exercise 7 - Generative Modelling

In this exercise you will be asked to apply some the principles of generative modelling you learned in the lecture to biological problems.

To learn a bit more about the theory of generative modelling, read through Chapter 17 (VAEs) and Chapter 18 (Diffusion Models) of the book Understanding Deep Learning. You do not need to grasp all the details, but try to understand the main ideas and the differences between the approaches.

After this, work through the following notebooks:

  1. Writing a Diffusion Encoder
  2. Training a 1D Diffusion Model
  3. Training a reparameterized 1D Diffusion Model. What is the main difference between the parametrization in this and the previous notebook?
  4. Train DDIM and accelerated models. What is the difference between the “classical” model from before and these two new ones?


The image was taken from the Stable Diffusion repository by StabilityAI.