Discriminating Against Unrealistic Interpolations in Generative Adversarial Networks

03/02/2022
by   Henning Petzka, et al.
0

Interpolations in the latent space of deep generative models is one of the standard tools to synthesize semantically meaningful mixtures of generated samples. As the generator function is non-linear, commonly used linear interpolations in the latent space do not yield the shortest paths in the sample space, resulting in non-smooth interpolations. Recent work has therefore equipped the latent space with a suitable metric to enforce shortest paths on the manifold of generated samples. These are often, however, susceptible of veering away from the manifold of real samples, resulting in smooth but unrealistic generation that requires an additional method to assess the sample quality along paths. Generative Adversarial Networks (GANs), by construction, measure the sample quality using its discriminator network. In this paper, we establish that the discriminator can be used effectively to avoid regions of low sample quality along shortest paths. By reusing the discriminator network to modify the metric on the latent space, we propose a lightweight solution for improved interpolations in pre-trained GANs.

READ FULL TEXT

page 13

page 14

page 15

page 16

page 17

page 22

page 23

page 24

research
07/18/2017

Optimizing the Latent Space of Generative Networks

Generative Adversarial Networks (GANs) have been shown to be able to sam...
research
11/03/2017

Metrics for Deep Generative Models

Neural samplers such as variational autoencoders (VAEs) or generative ad...
research
04/15/2019

Improved Precision and Recall Metric for Assessing Generative Models

The ability to evaluate the performance of a computational model is a vi...
research
10/19/2021

Latent reweighting, an almost free improvement for GANs

Standard formulations of GANs, where a continuous function deforms a con...
research
10/28/2020

GENs: Generative Encoding Networks

Mapping data from and/or onto a known family of distributions has become...
research
04/12/2018

Blood Vessel Geometry Synthesis using Generative Adversarial Networks

Computationally synthesized blood vessels can be used for training and e...
research
03/27/2021

Particle Filter Bridge Interpolation

Auto encoding models have been extensively studied in recent years. They...

Please sign up or login with your details

Forgot password? Click here to reset