Stacked convolutional auto-encoders for hierarchical feature extraction

Jonathan Masci; Ueli Meier; Dan Cireşan; Jürgen Schmidhuber

doi:10.1007/978-3-642-21735-7_7

Stacked convolutional auto-encoders for hierarchical feature extraction

Jonathan Masci, Ueli Meier, Dan Cireşan, Jürgen Schmidhuber

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

1661 Scopus citations

Abstract

We present a novel convolutional auto-encoder (CAE) for unsupervised feature learning. A stack of CAEs forms a convolutional neural network (CNN). Each CAE is trained using conventional on-line gradient descent without additional regularization terms. A max-pooling layer is essential to learn biologically plausible features consistent with those found by previous approaches. Initializing a CNN with filters of a trained CAE stack yields superior performance on a digit (MNIST) and an object recognition (CIFAR10) benchmark. © 2011 Springer-Verlag.

Original language	English (US)
Title of host publication	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages	52-59
Number of pages	8
DOIs	https://doi.org/10.1007/978-3-642-21735-7_7
State	Published - Jun 24 2011
Externally published	Yes

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-642-21735-7_7

Cite this

@inproceedings{653dade47d6740b2aa2ef26b5065a1d4,

title = "Stacked convolutional auto-encoders for hierarchical feature extraction",

abstract = "We present a novel convolutional auto-encoder (CAE) for unsupervised feature learning. A stack of CAEs forms a convolutional neural network (CNN). Each CAE is trained using conventional on-line gradient descent without additional regularization terms. A max-pooling layer is essential to learn biologically plausible features consistent with those found by previous approaches. Initializing a CNN with filters of a trained CAE stack yields superior performance on a digit (MNIST) and an object recognition (CIFAR10) benchmark. {\textcopyright} 2011 Springer-Verlag.",

author = "Jonathan Masci and Ueli Meier and Dan Cire{\c s}an and J{\"u}rgen Schmidhuber",

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

year = "2011",

month = jun,

day = "24",

doi = "10.1007/978-3-642-21735-7_7",

language = "English (US)",

isbn = "9783642217340",

pages = "52--59",

booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Stacked convolutional auto-encoders for hierarchical feature extraction

AU - Masci, Jonathan

AU - Meier, Ueli

AU - Cireşan, Dan

AU - Schmidhuber, Jürgen

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

PY - 2011/6/24

Y1 - 2011/6/24

N2 - We present a novel convolutional auto-encoder (CAE) for unsupervised feature learning. A stack of CAEs forms a convolutional neural network (CNN). Each CAE is trained using conventional on-line gradient descent without additional regularization terms. A max-pooling layer is essential to learn biologically plausible features consistent with those found by previous approaches. Initializing a CNN with filters of a trained CAE stack yields superior performance on a digit (MNIST) and an object recognition (CIFAR10) benchmark. © 2011 Springer-Verlag.

AB - We present a novel convolutional auto-encoder (CAE) for unsupervised feature learning. A stack of CAEs forms a convolutional neural network (CNN). Each CAE is trained using conventional on-line gradient descent without additional regularization terms. A max-pooling layer is essential to learn biologically plausible features consistent with those found by previous approaches. Initializing a CNN with filters of a trained CAE stack yields superior performance on a digit (MNIST) and an object recognition (CIFAR10) benchmark. © 2011 Springer-Verlag.

UR - http://link.springer.com/10.1007/978-3-642-21735-7_7

UR - http://www.scopus.com/inward/record.url?scp=79959353548&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-21735-7_7

DO - 10.1007/978-3-642-21735-7_7

M3 - Conference contribution

SN - 9783642217340

SP - 52

EP - 59

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

ER -

Stacked convolutional auto-encoders for hierarchical feature extraction

Abstract

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this