Log-Nets: Logarithmic Feature-Product Layers Yield More Compact Networks

Philipp Grüning; Thomas Martinetz; Erhardt Barth

doi:10.1007/978-3-030-61616-8_7

Log-Nets: Logarithmic Feature-Product Layers Yield More Compact Networks

Philipp Grüning^*, Thomas Martinetz, Erhardt Barth

^*Corresponding author for this work

2 Citations (Scopus)

Abstract

We introduce Logarithm-Networks (Log-Nets), a novel bio-inspired type of network architecture based on logarithms of feature maps followed by convolutions. Log-Nets are capable of surpassing the performance of traditional convolutional neural networks (CNNs) while using fewer parameters. Performance is evaluated on the Cifar-10 and ImageNet benchmarks.

Original language	English
Title of host publication	ICANN 2020: Artificial Neural Networks and Machine Learning – ICANN 2020
Editors	Igor Farkaš, Paolo Masulli, Stefan Wermter
Number of pages	13
Volume	12397 LNCS
Place of Publication	Cham
Publisher	Springer, Cham
Publication date	14.10.2020
Pages	79-91
ISBN (Print)	978-3-030-61615-1
ISBN (Electronic)	978-3-030-61616-8
DOIs	https://doi.org/10.1007/978-3-030-61616-8_7
Publication status	Published - 14.10.2020
Event	29th International Conference on Artificial Neural Networks - Bratislava, Slovakia Duration: 15.09.2020 → 18.09.2020 Conference number: 250349

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being
SDG 4 Quality Education
SDG 9 Industry, Innovation, and Infrastructure
SDG 11 Sustainable Cities and Communities
SDG 12 Responsible Consumption and Production
SDG 14 Life Below Water
SDG 15 Life on Land

Research Areas and Centers

Centers: Center for Artificial Intelligence Luebeck (ZKIL)
Research Area: Intelligent Systems

DFG Research Classification Scheme

4.43-05 Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing

Access to Document

10.1007/978-3-030-61616-8_7

Cite this

Grüning, P., Martinetz, T., & Barth, E. (2020). Log-Nets: Logarithmic Feature-Product Layers Yield More Compact Networks. In I. Farkaš, P. Masulli, & S. Wermter (Eds.), ICANN 2020: Artificial Neural Networks and Machine Learning – ICANN 2020 (Vol. 12397 LNCS, pp. 79-91). (Lecture Notes in Computer Science; Vol. 12397 LNCS). Springer, Cham. https://doi.org/10.1007/978-3-030-61616-8_7

@inproceedings{3ce0c2bb0260489189799d7ba636b1e8,

title = "Log-Nets: Logarithmic Feature-Product Layers Yield More Compact Networks",

abstract = "We introduce Logarithm-Networks (Log-Nets), a novel bio-inspired type of network architecture based on logarithms of feature maps followed by convolutions. Log-Nets are capable of surpassing the performance of traditional convolutional neural networks (CNNs) while using fewer parameters. Performance is evaluated on the Cifar-10 and ImageNet benchmarks.",

author = "Philipp Gr{\"u}ning and Thomas Martinetz and Erhardt Barth",

note = "Publisher Copyright: {\textcopyright} 2020, Springer Nature Switzerland AG. Copyright: Copyright 2020 Elsevier B.V., All rights reserved.; 29th International Conference on Artificial Neural Networks, ICANN 2020 ; Conference date: 15-09-2020 Through 18-09-2020",

year = "2020",

month = oct,

day = "14",

doi = "10.1007/978-3-030-61616-8\_7",

language = "English",

isbn = "978-3-030-61615-1",

volume = "12397 LNCS",

series = "Lecture Notes in Computer Science",

publisher = "Springer, Cham",

pages = "79--91",

editor = "Farka{\v s}, \{Igor \} and Masulli, \{Paolo \} and Wermter, \{Stefan \}",

booktitle = "ICANN 2020: Artificial Neural Networks and Machine Learning – ICANN 2020",

}

Grüning, P , Martinetz, T & Barth, E 2020, Log-Nets: Logarithmic Feature-Product Layers Yield More Compact Networks. in I Farkaš, P Masulli & S Wermter (eds), ICANN 2020: Artificial Neural Networks and Machine Learning – ICANN 2020. vol. 12397 LNCS, Lecture Notes in Computer Science, vol. 12397 LNCS, Springer, Cham, Cham, pp. 79-91, 29th International Conference on Artificial Neural Networks, Bratislava, Slovakia, 15.09.20. https://doi.org/10.1007/978-3-030-61616-8_7

Log-Nets: Logarithmic Feature-Product Layers Yield More Compact Networks. / Grüning, Philipp ; Martinetz, Thomas ; Barth, Erhardt.
ICANN 2020: Artificial Neural Networks and Machine Learning – ICANN 2020. ed. / Igor Farkaš; Paolo Masulli; Stefan Wermter. Vol. 12397 LNCS Cham: Springer, Cham, 2020. p. 79-91 (Lecture Notes in Computer Science; Vol. 12397 LNCS).

Research output: Chapters in Books/Reports/Conference Proceedings › Conference contribution › peer-review

TY - GEN

T1 - Log-Nets: Logarithmic Feature-Product Layers Yield More Compact Networks

AU - Grüning, Philipp

AU - Martinetz, Thomas

AU - Barth, Erhardt

N1 - Conference code: 250349

PY - 2020/10/14

Y1 - 2020/10/14

N2 - We introduce Logarithm-Networks (Log-Nets), a novel bio-inspired type of network architecture based on logarithms of feature maps followed by convolutions. Log-Nets are capable of surpassing the performance of traditional convolutional neural networks (CNNs) while using fewer parameters. Performance is evaluated on the Cifar-10 and ImageNet benchmarks.

AB - We introduce Logarithm-Networks (Log-Nets), a novel bio-inspired type of network architecture based on logarithms of feature maps followed by convolutions. Log-Nets are capable of surpassing the performance of traditional convolutional neural networks (CNNs) while using fewer parameters. Performance is evaluated on the Cifar-10 and ImageNet benchmarks.

UR - https://www.scopus.com/pages/publications/85094177891

U2 - 10.1007/978-3-030-61616-8_7

DO - 10.1007/978-3-030-61616-8_7

M3 - Conference contribution

SN - 978-3-030-61615-1

VL - 12397 LNCS

T3 - Lecture Notes in Computer Science

SP - 79

EP - 91

BT - ICANN 2020: Artificial Neural Networks and Machine Learning – ICANN 2020

A2 - Farkaš, Igor

A2 - Masulli, Paolo

A2 - Wermter, Stefan

PB - Springer, Cham

CY - Cham

T2 - 29th International Conference on Artificial Neural Networks

Y2 - 15 September 2020 through 18 September 2020

ER -

Log-Nets: Logarithmic Feature-Product Layers Yield More Compact Networks

Abstract

UN SDGs

Research Areas and Centers

DFG Research Classification Scheme

Access to Document

Other files and links

Fingerprint

Cite this