CHAPTER 15 / 40
Generative Models
5 lessons · Explore, experiment and understand.
Let’s figure it out.
Ask me about this course or where to start.
- Step 66~5mCode · 11🎯Quick win
Variational Autoencoders & The Reparameterization Trick
A VAE is a generative model that learns to compress data into a continuous latent space and decode latents back into data. Two networks: an…
Open lesson - Step 67~5m3D · 1Code · 11★Most cinematic
GANs and the Wasserstein Distance
A GAN (Goodfellow et al. 2014) pits two networks against each other: a generator maps noise to fake samples; a discriminator tries to tell r…
Open lesson - Step 68~5mCode · 11🎯Quick win
Normalizing Flows: Invertible Densities
A normalizing flow is a sequence of invertible neural transformations that map a simple base distribution (usually Gaussian) into a complex…
Open lesson - Step 69~5mCode · 11🎯Quick win
Autoregressive Generative Models
An autoregressive model factorizes a joint distribution as a product of conditionals: . Each is generated sequentially given all previous ou…
Open lesson - Step 70~5m3D · 1Code · 11★Most cinematic
Diffusion Models: Forward and Reverse Process
A diffusion model generates data by reversing a gradual noise-adding process. The forward process corrupts a data point over steps: where .…
Open lesson