Wasserstein Auto-Encoders: Difference between revisions
Jump to navigation
Jump to search
| Line 1: | Line 1: | ||
= Introduction = | = Introduction = | ||
Recent years have seen a convergence of two previously distinct approaches: representation learning from high dimensional data, and unsupervised generative modeling. In the field that formed at the intersection, Variational Auto-Encoders (VAEs) and Generative Adversarial Networks (GANs) have emerged to be the most popular. VAEs are theoretically elegant but with the drawback that they tend to generate blurry samples when applied to natural images. GANs on the other hand produce better visual quality of sampled images, but come without an encoder, are harder to train and suffer from mode-collapse problem. | Recent years have seen a convergence of two previously distinct approaches: representation learning from high dimensional data, and unsupervised generative modeling. In the field that formed at the intersection, Variational Auto-Encoders (VAEs) and Generative Adversarial Networks (GANs) have emerged to be the most popular. VAEs are theoretically elegant but with the drawback that they tend to generate blurry samples when applied to natural images. GANs on the other hand produce better visual quality of sampled images, but come without an encoder, are harder to train and suffer from the mode-collapse problem when the trained model is unable to capture all the variability in the true data distribution. | ||
= Motivation = | = Motivation = | ||
Revision as of 21:43, 11 March 2018
Introduction
Recent years have seen a convergence of two previously distinct approaches: representation learning from high dimensional data, and unsupervised generative modeling. In the field that formed at the intersection, Variational Auto-Encoders (VAEs) and Generative Adversarial Networks (GANs) have emerged to be the most popular. VAEs are theoretically elegant but with the drawback that they tend to generate blurry samples when applied to natural images. GANs on the other hand produce better visual quality of sampled images, but come without an encoder, are harder to train and suffer from the mode-collapse problem when the trained model is unable to capture all the variability in the true data distribution.