Symmetric Variational Autoencoder and Connections to Adversarial Learning

2017/09/06 by Li‐Qun Chen, Chen, Liqun, Shuyang Dai +9 · 4 citations
Computer Science · #Generative Adversarial Networks and Image Synthesis #Adversarial Robustness in Machine Learning #Anomaly Detection Techniques and Applications

paper · pdf · doi:10.48550/arxiv.1709.01846

Abstract

A new form of the variational autoencoder (VAE) is proposed, based on the symmetric Kullback-Leibler divergence. It is demonstrated that learning of the resulting symmetric VAE (sVAE) has close connections to previously developed adversarial-learning methods. This relationship helps unify the previously distinct techniques of VAE and adversarially learning, and provides insights that allow us to ameliorate shortcomings with some previously developed adversarial methods. In addition to an analysis that motivates and explains the sVAE, an extensive set of experiments validate the utility of the approach.

Citations

• Auto-Encoding Variational Bayes
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks

Cited by

• Lifelong Mixture of Variational Autoencoders
• Tutorial: Deriving the Standard Variational Autoencoder (VAE) Loss Function
• Learning latent representations across multiple data domains using Lifelong VAEGAN
• Adversarial Symmetric Variational Autoencoder

Related

• Adversarial Symmetric Variational Autoencoder
• On the Importance of the Kullback-Leibler Divergence Term in Variational Autoencoders for Text Generation
• Symmetric Equilibrium Learning of VAEs
• How do Variational Autoencoders Learn? Insights from Representational Similarity
• Tutorial on Variational Autoencoders