Writing
From Probabilistic Modeling to Generative Modeling: a series that starts from probability rules and maximum likelihood, then works through Gaussian mixtures, VAEs, normalizing flows, GANs and diffusion models, covering each model's motivation, derivations and limitations. Best read in order.
-
Part 01 Probabilistic Generative Models Overview An introduction to probabilistic generative modeling: how generative models differ from discriminative ones, and the concepts behind five key model families. -
Part 02 Gaussian Mixture Models Explained Gaussian Mixture Models express a probability distribution as a weighted combination of Gaussians, capturing multi-modal data through interpretable components. -
Part 03 Variational Autoencoders Explained A Variational Autoencoder (VAE) is a generative model that learns a compressed, continuous representation (a latent space) of data. It consists of an encoder network that maps data to a distribution in the latent space and a decoder network that reconstructs data from samples drawn from that latent distribution. -
Part 04 Normalizing Flows Explained Normalizing Flows turn a simple Gaussian into a complex distribution through invertible transformations, enabling exact and tractable likelihood computation. -
Part 05 Generative Adversarial Networks Explained How GANs learn by pitting a generator against a discriminator: the minimax game, and why adversarial training replaced explicit density estimation. -
Part 06 Diffusion Models Explained How diffusion models generate images by reversing a gradual noise process: forward and reverse Markov chains, DDPM, and why they rival GANs on sample quality.