Nonlinear ICA through low-complexity autoencoders

Sepp Hochreiter; Juergen Schmidhuber

Nonlinear ICA through low-complexity autoencoders

Research output: Contribution to journal › Article › peer-review

Abstract

Flat Minimum Search, a regularizer algorithm for finding low-complexity networks describable by few bits of information, is employed to train autocoders. Flat minima are regions in weight space where there is tolerable small error and the weights can be perturbed without greatly affecting the network's output. The procedure reveals an important connection between regularization and independent component analysis. This connection may represent a first step towards unification of regularization and unsupervised learning.

Original language	English (US)
Journal	Proceedings - IEEE International Symposium on Circuits and Systems
Volume	5
State	Published - Jan 1 1999
Externally published	Yes

Cite this

@article{4db7dc8842ae477ea5154391ba5dfceb,

title = "Nonlinear ICA through low-complexity autoencoders",

abstract = "Flat Minimum Search, a regularizer algorithm for finding low-complexity networks describable by few bits of information, is employed to train autocoders. Flat minima are regions in weight space where there is tolerable small error and the weights can be perturbed without greatly affecting the network's output. The procedure reveals an important connection between regularization and independent component analysis. This connection may represent a first step towards unification of regularization and unsupervised learning.",

author = "Sepp Hochreiter and Juergen Schmidhuber",

note = "Generated from Scopus record by KAUST IRTS on 2022-09-14",

year = "1999",

month = jan,

day = "1",

language = "English (US)",

volume = "5",

journal = "Proceedings - IEEE International Symposium on Circuits and Systems",

issn = "0271-4310",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - JOUR

T1 - Nonlinear ICA through low-complexity autoencoders

AU - Hochreiter, Sepp

AU - Schmidhuber, Juergen

N1 - Generated from Scopus record by KAUST IRTS on 2022-09-14

PY - 1999/1/1

Y1 - 1999/1/1

N2 - Flat Minimum Search, a regularizer algorithm for finding low-complexity networks describable by few bits of information, is employed to train autocoders. Flat minima are regions in weight space where there is tolerable small error and the weights can be perturbed without greatly affecting the network's output. The procedure reveals an important connection between regularization and independent component analysis. This connection may represent a first step towards unification of regularization and unsupervised learning.

AB - Flat Minimum Search, a regularizer algorithm for finding low-complexity networks describable by few bits of information, is employed to train autocoders. Flat minima are regions in weight space where there is tolerable small error and the weights can be perturbed without greatly affecting the network's output. The procedure reveals an important connection between regularization and independent component analysis. This connection may represent a first step towards unification of regularization and unsupervised learning.

UR - http://www.scopus.com/inward/record.url?scp=0032662583&partnerID=8YFLogxK

M3 - Article

SN - 0271-4310

VL - 5

JO - Proceedings - IEEE International Symposium on Circuits and Systems

JF - Proceedings - IEEE International Symposium on Circuits and Systems

ER -

Nonlinear ICA through low-complexity autoencoders

Abstract

Other files and links

Fingerprint

Cite this