Thumbnail

X-Ray Sobolev Variational Auto-Encoders

Turinici, Gabriel (2019), X-Ray Sobolev Variational Auto-Encoders. https://basepub.dauphine.fr/handle/123456789/20349

View/Open
Type
Document de travail / Working paper
External document link
https://hal.archives-ouvertes.fr/hal-02387084
Date
2019
Publisher
Cahier de recherche CEREMADE, Université Paris-Dauphine
Series title
Cahier de recherche CEREMADE, Université Paris-Dauphine
Published in
Paris
Pages
26
Metadata
Show full item record
Author(s)
Turinici, Gabriel cc
CEntre de REcherches en MAthématiques de la DEcision [CEREMADE]
Abstract (EN)
The quality of the generative models (Generative adversarial networks, Variational Auto-Encoders, ...) depends heavily on the choice of a good probability distance. However some popular metrics lack convenient properties such as (geodesic) convexity, fast evaluation and so on. To address these shortcomings, we introduce a class of distances that have built-in convexity. We investigate the relationship with some known paradigms (sliced distances, reproducing kernel Hilbert spaces, energy distances). The distances are shown to posses fast implementations and are included in an adapted Variational Auto-Encoder termed X-ray Sobolev Variational Auto-Encoder (XS-VAE) which produces good quality results on standard generative datasets.
Subjects / Keywords
machine learning; variational auto-encoder; neural network; deep learning