XCS236: Deep Generative Models
Stanford School of Engineering
Course Overview
A course on deep generative models covering probabilistic foundations and learning algorithms for various model types.
Topics Covered
- Autoregressive Models
- Variational Autoencoders
- Generative Adversarial Networks
- Flow-based Models
- Energy-based Models
- Diffusion Models
My Course Notes
Module Notes
Module 1: Introduction to Generative Models
Fundamentals of generative models, probability distributions, and maximum likelihood estimation.
View NotesModule 2: Autoregressive Models
Sequential data generation models including NADE, PixelRNN, and PixelCNN.
View NotesModule 3: Variational Autoencoders
Latent variable models using variational inference and the reparameterization trick.
View NotesModule 4: Generative Adversarial Networks
Adversarial training approach with generator and discriminator networks.
View NotesModule 5: Advanced Topics in Generative Models
Flow-based models, energy-based models, and diffusion models.
View Notes