Lagging Inference Networks and Posterior Collapse in Variational Autoencoders
ICLR, Volume abs/1901.05534, 2019.
The variational autoencoder (VAE) is a popular combination of deep latent variable model and accompanying variational learning technique. By using a neural inference network to approximate the modelu0027s on latent variables, VAEs efficiently parameterize a lower bound on marginal data likelihood that can be optimized directly via gradie...More
PPT (Upload PPT)