Lagging Inference Networks and Posterior Collapse in Variational Autoencoders

ICLR 2019 Junxian HeDaniel SpokoynyGraham NeubigTaylor Berg-Kirkpatrick

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 model's posterior on latent variables, VAEs efficiently parameterize a lower bound on marginal data likelihood that can be optimized directly via gradient methods... (read more)

PDF Abstract

Evaluation results from the paper


Task Dataset Model Metric name Metric value Global rank Compare
Text Generation Yahoo Questions Aggressive VAE NLL 326.7 # 1
Text Generation Yahoo Questions Aggressive VAE KL 5.7 # 3
Text Generation Yahoo Questions Aggressive VAE Perplexity 59.7 # 1