On the expressivity of bi-Lipschitz normalizing flows

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Verine, Alexandre, Negrevergne, Benjamin, Rossi, Fabrice, Chevaleyre, Yann
Format: Preprint
Published: 2021
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929267370098688
author Verine, Alexandre
Negrevergne, Benjamin
Rossi, Fabrice
Chevaleyre, Yann
author_facet Verine, Alexandre
Negrevergne, Benjamin
Rossi, Fabrice
Chevaleyre, Yann
contents An invertible function is bi-Lipschitz if both the function and its inverse have bounded Lipschitz constants. Nowadays, most Normalizing Flows are bi-Lipschitz by design or by training to limit numerical errors (among other things). In this paper, we discuss the expressivity of bi-Lipschitz Normalizing Flows and identify several target distributions that are difficult to approximate using such models. Then, we characterize the expressivity of bi-Lipschitz Normalizing Flows by giving several lower bounds on the Total Variation distance between these particularly unfavorable distributions and their best possible approximation. Finally, we discuss potential remedies which include using more complex latent distributions.
format Preprint
id arxiv_https___arxiv_org_abs_2107_07232
institution arXiv
publishDate 2021
record_format arxiv
spellingShingle On the expressivity of bi-Lipschitz normalizing flows
Verine, Alexandre
Negrevergne, Benjamin
Rossi, Fabrice
Chevaleyre, Yann
Machine Learning
An invertible function is bi-Lipschitz if both the function and its inverse have bounded Lipschitz constants. Nowadays, most Normalizing Flows are bi-Lipschitz by design or by training to limit numerical errors (among other things). In this paper, we discuss the expressivity of bi-Lipschitz Normalizing Flows and identify several target distributions that are difficult to approximate using such models. Then, we characterize the expressivity of bi-Lipschitz Normalizing Flows by giving several lower bounds on the Total Variation distance between these particularly unfavorable distributions and their best possible approximation. Finally, we discuss potential remedies which include using more complex latent distributions.
title On the expressivity of bi-Lipschitz normalizing flows
topic Machine Learning
url https://arxiv.org/abs/2107.07232