Generalization Properties of Optimal Transport GANs with Latent Distribution Learning

07/29/2020
by   Giulia Luise, et al.
0

The Generative Adversarial Networks (GAN) framework is a well-established paradigm for probability matching and realistic sample generation. While recent attention has been devoted to studying the theoretical properties of such models, a full theoretical understanding of the main building blocks is still missing. Focusing on generative models with Optimal Transport metrics as discriminators, in this work we study how the interplay between the latent distribution and the complexity of the pushforward map (generator) affects performance, from both statistical and modelling perspectives. Motivated by our analysis, we advocate learning the latent distribution as well as the pushforward map within the GAN paradigm. We prove that this can lead to significant advantages in terms of sample complexity.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
06/24/2019

Adversarial Computation of Optimal Transport Maps

Computing optimal transport maps between high-dimensional and continuous...
research
02/14/2022

Understanding DDPM Latent Codes Through Optimal Transport

Diffusion models have recently outperformed alternative approaches to mo...
research
01/11/2020

AE-OT-GAN: Training GANs from data specific latent distribution

Though generative adversarial networks (GANs) areprominent models to gen...
research
12/04/2019

Informative GANs via Structured Regularization of Optimal Transport

We tackle the challenge of disentangled representation learning in gener...
research
01/14/2021

Convex Smoothed Autoencoder-Optimal Transport model

Generative modelling is a key tool in unsupervised machine learning whic...
research
10/29/2021

Learning generative models for valid knockoffs using novel multivariate-rank based statistics

We consider the problem of generating valid knockoffs for knockoff filte...
research
08/29/2019

Potential Flow Generator with L_2 Optimal Transport Regularity for Generative Models

We propose a potential flow generator with L_2 optimal transport regular...

Please sign up or login with your details

Forgot password? Click here to reset