Lagging Inference Networks and Posterior Collapse in Variational Autoencoders.

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 gradient methods. In practice, however, VAE training often results in a degenerate local optimum known as posterior where the model learns to ignore the latent variable and the approximate mimics the prior. In this paper, we investigate collapse from the perspective of training dynamics. We find that during the initial stages of training the inference network fails to approximate the modelu0027s true posterior, which is a moving target. As a result, the model is encouraged to ignore the latent encoding and collapse occurs. Based on this observation, we propose an extremely simple modification to VAE training to reduce inference lag: depending on the modelu0027s current mutual information between latent variable and observation, we aggressively optimize the inference network before performing each model update. Despite introducing neither new model components nor significant complexity over basic VAE, our approach is able to avoid the problem of collapse that has plagued a large amount of previous work. Empirically, our approach outperforms strong autoregressive baselines on text and image benchmarks in terms of held-out likelihood, and is competitive with more complex techniques for avoiding collapse while being substantially faster.