Variational Autoencoder - Graph View

A generative model that learns a structured, continuous latent space by combining autoencoder architecture with probabilistic inference, enabling generation of new data by sampling from the learned distribution.