Learnable Explicit Density for Continuous Latent Space and Variational Inference

Abstract

In this paper, we study two aspects of the variational autoencoder (VAE): the prior distribution over the latent variables and its corresponding posterior. First, we decompose the learning of VAEs into layerwise density estimation, and argue that having a flexible prior is beneficial to both sample generation and inference. Second, we analyze the family of inverse autoregressive flows (inverse AF) and show that with further improvement, inverse AF could be used as universal approximation to any complicated posterior. Our analysis results in a unified approach to parameterizing a VAE, without the need to restrict ourselves to use factorial Gaussians in the latent real space.

2 figures, 5 pages, submitted to ICML Principled Approaches to Deep Learning workshop

References in corpus (6)

• Auto-Encoding Variational Bayes
• Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)
• Variational Inference with Normalizing Flows
• Density estimation using Real NVP
• MADE: Masked Autoencoder for Distribution Estimation
• Improving Variational Inference with Inverse Autoregressive Flow

Cited by in corpus (10)

• Deep Generative Modelling: A Comparative Review of VAEs, GANs, Normalizing Flows, Energy-Based and Autoregressive Models
• Very Deep VAEs Generalize Autoregressive Models and Can Outperform Them on Images
• A Tutorial on Deep Latent Variable Models of Natural Language
• Augmented Normalizing Flows: Bridging the Gap Between Generative Flows and Latent Variable Models
• Generative Latent Flow
• Variational Autoencoders with Normalizing Flow Decoders
• On the Necessity and Effectiveness of Learning the Prior of Variational Auto-Encoder
• Improving Sequential Latent Variable Models with Autoregressive Flows
• Benefiting Deep Latent Variable Models via Learning the Prior and Removing Latent Regularization
• To Regularize or Not To Regularize? The Bias Variance Trade-off in Regularized AEs