Bøker av Diederik P. Kingma

Presents an introduction to the framework of variational autoencoders (VAEs) that provides a principled method for jointly learning deep latent-variable models and corresponding inference models using stochastic gradient descent.