End-to-End Learning of Geometrical Shaping Maximizing Generalized Mutual Information

Kadir Gumus; Alex Alvarado; Bin Chen; Christian Hager; Erik Agrell

doi:10.1364/OFC-2020-W3D.4

End-to-End Learning of Geometrical Shaping Maximizing Generalized Mutual Information

Kadir Gumus, Alex Alvarado, Bin Chen, Christian Hager, Erik Agrell

Onderzoeksoutput: Hoofdstuk in Boek/Rapport/Congresprocedure › Conferentiebijdrage › Academic › peer review

57 Citaties (SciVal)

97 Downloads (Pure)

Samenvatting

GMI-based end-to-end learning is shown to be highly nonconvex. We apply gradient descent initialized with Gray-labeled APSK constellations directly to the constellation coordinates. State-of-the-art constellations in 2D and 4D are found providing reach increases up to 26% w.r.t. to QAM.

Originele taal-2	Engels
Titel	2020 Optical Fiber Communications Conference and Exhibition, OFC 2020 - Proceedings
Uitgeverij	Institute of Electrical and Electronics Engineers
Hoofdstuk	3
ISBN van elektronische versie	9781943580712
ISBN van geprinte versie	9781943580712
DOI's	https://doi.org/10.1364/OFC-2020-W3D.4
Status	Gepubliceerd - 13 mei 2020
Evenement	The Optical Networking and Communication Conference & Exhibition (OFC2020) - San Diego, Verenigde Staten van Amerika Duur: 8 mrt. 2020 → 12 mrt. 2020 https://www.ofcconference.org/en-us/home/

Congres

Congres	The Optical Networking and Communication Conference & Exhibition (OFC2020)
Verkorte titel	OFC2020
Land/Regio	Verenigde Staten van Amerika
Stad	San Diego
Periode	8/03/20 → 12/03/20
Internet adres	https://www.ofcconference.org/en-us/home/

Financiering

Acknowledgements: This work was funded by the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant no. 749798 and the ERC grant no. 757791, the National Natural Science Foundation of China (NSFC) under grant no. 61701155, the Fundamental Research Funds for the Central Universities under grant no. JZ2019HGBZ0130, the Netherlands Organisation for Scientific Research (NWO) via the VIDI project ICONIC under grant no. 15685, and the Swedish Research Council under grant no. 2017-03702.

Financiers	Financiernummer
European Union’s Horizon Europe research and innovation programme	757791
H2020 Marie Skłodowska-Curie Actions	749798
European Research Council
National Natural Science Foundation of China	61701155
Nederlandse Organisatie voor Wetenschappelijk Onderzoek	15685
Vetenskapsrådet	2017-03702
Fundamental Research Funds for the Central Universities	JZ2019HGBZ0130

Toegang tot document

10.1364/OFC-2020-W3D.4

Author's versionFinale auteursversie , 235 KB

https://ieeexplore.ieee.org/document/9083181

Andere bestanden en links

Link to publication in Scopus

Fundamentals of the Nonlinear Optical Channel
Alvarado, A. (Project Manager), Liga, G. (Projectmedewerker), Barreiro, A. (Projectmedewerker), Willems, F. M. J. (Project communicatie medewerker), Sanders, R. (Project communicatie medewerker), Alvarado, A. (Project communicatie medewerker), Barreiro, A. (Project communicatie medewerker), Sheikh, A. (Projectmedewerker), Goossens, S. (Projectmedewerker), de Jonge, M. (Project communicatie medewerker), Gültekin, Y. C. (Projectmedewerker), Jaffal, Y. (Projectmedewerker), Oliari, V. (Projectmedewerker) & Ramachandran, V. (Projectmedewerker)
1/01/18 → 30/06/23
Project: Onderzoek direct
ICONIC: Increasing the Capacity of Optical Nonlinear Interfering Channels
Alvarado, A. (Project Manager), Alvarado, A. (Projectmedewerker), Willems, F. M. J. (Project communicatie medewerker), Sanders, R. (Project communicatie medewerker), Alvarado, A. (Project communicatie medewerker), Barreiro, A. (Project communicatie medewerker), Wu, K. (Projectmedewerker), de Jonge, M. (Project communicatie medewerker), Karanov, B. (Project communicatie medewerker), Karanov, B. (Projectmedewerker), Lee, J. (Projectmedewerker) & Oliari, V. (Projectmedewerker)
1/08/17 → 31/07/23
Project: Onderzoek direct

Citeer dit

@inproceedings{74ddf4ca15794c69af83f520e03179bc,

title = "End-to-End Learning of Geometrical Shaping Maximizing Generalized Mutual Information",

abstract = "GMI-based end-to-end learning is shown to be highly nonconvex. We apply gradient descent initialized with Gray-labeled APSK constellations directly to the constellation coordinates. State-of-the-art constellations in 2D and 4D are found providing reach increases up to 26% w.r.t. to QAM.",

author = "Kadir Gumus and Alex Alvarado and Bin Chen and Christian Hager and Erik Agrell",

year = "2020",

month = may,

day = "13",

doi = "10.1364/OFC-2020-W3D.4",

language = "English",

isbn = "9781943580712",

booktitle = "2020 Optical Fiber Communications Conference and Exhibition, OFC 2020 - Proceedings",

publisher = "Institute of Electrical and Electronics Engineers",

address = "United States",

note = "2020 Optical Fiber Communications Conference and Exhibition, OFC 2020, OFC 2020 ; Conference date: 08-03-2020 Through 12-03-2020",

url = "https://www.ofcconference.org/en-us/home/",

}

Gumus, K , Alvarado, A , Chen, B, Hager, C & Agrell, E 2020, End-to-End Learning of Geometrical Shaping Maximizing Generalized Mutual Information. in 2020 Optical Fiber Communications Conference and Exhibition, OFC 2020 - Proceedings., 9083181, Institute of Electrical and Electronics Engineers, The Optical Networking and Communication Conference & Exhibition (OFC2020), San Diego, California, Verenigde Staten van Amerika, 8/03/20. https://doi.org/10.1364/OFC-2020-W3D.4

End-to-End Learning of Geometrical Shaping Maximizing Generalized Mutual Information. / Gumus, Kadir ; Alvarado, Alex ; Chen, Bin et al.
2020 Optical Fiber Communications Conference and Exhibition, OFC 2020 - Proceedings. Institute of Electrical and Electronics Engineers, 2020. 9083181.

Onderzoeksoutput: Hoofdstuk in Boek/Rapport/Congresprocedure › Conferentiebijdrage › Academic › peer review

TY - GEN

T1 - End-to-End Learning of Geometrical Shaping Maximizing Generalized Mutual Information

AU - Gumus, Kadir

AU - Alvarado, Alex

AU - Chen, Bin

AU - Hager, Christian

AU - Agrell, Erik

PY - 2020/5/13

Y1 - 2020/5/13

N2 - GMI-based end-to-end learning is shown to be highly nonconvex. We apply gradient descent initialized with Gray-labeled APSK constellations directly to the constellation coordinates. State-of-the-art constellations in 2D and 4D are found providing reach increases up to 26% w.r.t. to QAM.

AB - GMI-based end-to-end learning is shown to be highly nonconvex. We apply gradient descent initialized with Gray-labeled APSK constellations directly to the constellation coordinates. State-of-the-art constellations in 2D and 4D are found providing reach increases up to 26% w.r.t. to QAM.

UR - http://www.scopus.com/inward/record.url?scp=85085176983&partnerID=8YFLogxK

U2 - 10.1364/OFC-2020-W3D.4

DO - 10.1364/OFC-2020-W3D.4

M3 - Conference contribution

AN - SCOPUS:85085176983

SN - 9781943580712

BT - 2020 Optical Fiber Communications Conference and Exhibition, OFC 2020 - Proceedings

PB - Institute of Electrical and Electronics Engineers

T2 - 2020 Optical Fiber Communications Conference and Exhibition, OFC 2020

Y2 - 8 March 2020 through 12 March 2020

ER -

End-to-End Learning of Geometrical Shaping Maximizing Generalized Mutual Information

Samenvatting

Congres

Financiering

Toegang tot document

Andere bestanden en links

Vingerafdruk

Projecten

Fundamentals of the Nonlinear Optical Channel

ICONIC: Increasing the Capacity of Optical Nonlinear Interfering Channels

Citeer dit