Flow-based generative model

Flow-based generative models are a class of probabilistic models that utilize invertible transformations to model complex distributions. These models are designed to generate new data samples from a learned distribution by applying a sequence of transformations to a simple base distribution, typically a multivariate Gaussian.