Ensuring Topological Data-Structure Preservation under Autoencoder Compression Due to Latent Space R...
Ensuring Topological Data-Structure Preservation under Autoencoder Compression Due to Latent Space Regularization in Gauss–Legendre Nodes
About this item
Full title
Author / Creator
Publisher
Basel: MDPI AG
Journal title
Language
English
Formats
Publication information
Publisher
Basel: MDPI AG
Subjects
More information
Scope and Contents
Contents
We formulate a data-independent latent space regularization constraint for general unsupervised autoencoders. The regularization relies on sampling the autoencoder Jacobian at Legendre nodes, which are the centers of the Gauss–Legendre quadrature. Revisiting this classic allows us to prove that regularized autoencoders ensure a one-to-one re-embedd...
Alternative Titles
Full title
Ensuring Topological Data-Structure Preservation under Autoencoder Compression Due to Latent Space Regularization in Gauss–Legendre Nodes
Authors, Artists and Contributors
Identifiers
Primary Identifiers
Record Identifier
TN_cdi_doaj_primary_oai_doaj_org_article_7b0e9c2a7855429c89a565acb2ed510a
Permalink
https://devfeature-collection.sl.nsw.gov.au/record/TN_cdi_doaj_primary_oai_doaj_org_article_7b0e9c2a7855429c89a565acb2ed510a
Other Identifiers
ISSN
2075-1680
E-ISSN
2075-1680
DOI
10.3390/axioms13080535
