Variational Autoencoder

machine learninggenerative modelingBayesianneural networksprobabilitycetztikz

Variational autoencoder (VAE) architecture. A foundational type of generative machine learning model that learns a latent representation by encoding inputs into a distribution and sampling from it. Inspired by https://towardsdatascience.com/intuitively-understanding-variational-autoencoders-1bfe67eb5daf.