What is the difference between Auto-encoding Variational Bayes and Stochastic Backpropagation for Deep Generative Models? Does inference in both methods lead to the same results? I'm not aware of any explicit comparisons between the two methods, despite that both groups of authors cite each other.
Stochastic Backpropagation and Approximate Inference in Deep Generative Models seems to just consider a slightly more general implementation of the same ideas. Compared to the VAE paper, they:
propose multiple sets of latent variables, as opposed to VAE which has just one
consider non-diagonal gaussian posteriors, which make it more difficult to optimize