Academic Bibliography

 Search 200 years of publications by Ghent University researchers.
Advanced search
1 file | 1.68 MB
Add to list
  1. Search results
  2. Comparing different nonlinearities in...

Comparing different nonlinearities in readout systems for optical neuromorphic computing networks

Chonghuai Ma, Joris Lambrecht (UGent) , Floris Laporte, Xin Yin (UGent) , Joni Dambre (UGent) and Peter Bienstman (UGent)
(2021) SCIENTIFIC REPORTS. 11(1).
Author
Organization
Project
Abstract
Nonlinear activation is a crucial building block of most machine-learning systems. However, unlike in the digital electrical domain, applying a saturating nonlinear function in a neural network in the analog optical domain is not as easy, especially in integrated systems. In this paper, we first investigate in detail the photodetector nonlinearity in two main readout schemes: electrical readout and optical readout. On a 3-bit-delayed XOR task, we show that optical readout trained with backpropagation gives the best performance. Furthermore, we propose an additional saturating nonlinearity coming from a deliberately non-ideal voltage amplifier after the detector. Compared to an all-optical nonlinearity, these two kinds of nonlinearities are extremely easy to obtain at no additional cost, since photodiodes and voltage amplifiers are present in any system. Moreover, not having to design ideal linear amplifiers could relax their design requirements. We show through simulation that for long-distance nonlinear fiber distortion compensation, using only the photodiode nonlinearity in an optical readout delivers BER improvements over three orders of magnitude. Combined with the amplifier saturation nonlinearity, we obtain another three orders of magnitude improvement of the BER.

Downloads

  • Download
    pub 2735.pdf
    • full text (Published version)
    • |
    • open access
    • |
    • PDF
    • |
    • 1.68 MB

Citation

Please use this url to cite or link to this publication:

MLA
Ma, Chonghuai, et al. “Comparing Different Nonlinearities in Readout Systems for Optical Neuromorphic Computing Networks.” SCIENTIFIC REPORTS, vol. 11, no. 1, 2021, doi:10.1038/s41598-021-03594-0.
APA
Ma, C., Lambrecht, J., Laporte, F., Yin, X., Dambre, J., & Bienstman, P. (2021). Comparing different nonlinearities in readout systems for optical neuromorphic computing networks. SCIENTIFIC REPORTS, 11(1). https://doi.org/10.1038/s41598-021-03594-0
Chicago author-date
Ma, Chonghuai, Joris Lambrecht, Floris Laporte, Xin Yin, Joni Dambre, and Peter Bienstman. 2021. “Comparing Different Nonlinearities in Readout Systems for Optical Neuromorphic Computing Networks.” SCIENTIFIC REPORTS 11 (1). https://doi.org/10.1038/s41598-021-03594-0.
Chicago author-date (all authors)
Ma, Chonghuai, Joris Lambrecht, Floris Laporte, Xin Yin, Joni Dambre, and Peter Bienstman. 2021. “Comparing Different Nonlinearities in Readout Systems for Optical Neuromorphic Computing Networks.” SCIENTIFIC REPORTS 11 (1). doi:10.1038/s41598-021-03594-0.
Vancouver
1.
Ma C, Lambrecht J, Laporte F, Yin X, Dambre J, Bienstman P. Comparing different nonlinearities in readout systems for optical neuromorphic computing networks. SCIENTIFIC REPORTS. 2021;11(1).
IEEE
[1]
C. Ma, J. Lambrecht, F. Laporte, X. Yin, J. Dambre, and P. Bienstman, “Comparing different nonlinearities in readout systems for optical neuromorphic computing networks,” SCIENTIFIC REPORTS, vol. 11, no. 1, 2021.
@article{8738675,
  abstract     = {{Nonlinear activation is a crucial building block of most machine-learning systems. However, unlike in the digital electrical domain, applying a saturating nonlinear function in a neural network in the analog optical domain is not as easy, especially in integrated systems. In this paper, we first investigate in detail the photodetector nonlinearity in two main readout schemes: electrical readout and optical readout. On a 3-bit-delayed XOR task, we show that optical readout trained with backpropagation gives the best performance. Furthermore, we propose an additional saturating nonlinearity coming from a deliberately non-ideal voltage amplifier after the detector. Compared to an all-optical nonlinearity, these two kinds of nonlinearities are extremely easy to obtain at no additional cost, since photodiodes and voltage amplifiers are present in any system. Moreover, not having to design ideal linear amplifiers could relax their design requirements. We show through simulation that for long-distance nonlinear fiber distortion compensation, using only the photodiode nonlinearity in an optical readout delivers BER improvements over three orders of magnitude. Combined with the amplifier saturation nonlinearity, we obtain another three orders of magnitude improvement of the BER.}},
  articleno    = {{24152}},
  author       = {{Ma, Chonghuai and Lambrecht, Joris and Laporte, Floris and Yin, Xin and Dambre, Joni and Bienstman, Peter}},
  issn         = {{2045-2322}},
  journal      = {{SCIENTIFIC REPORTS}},
  language     = {{eng}},
  number       = {{1}},
  pages        = {{8}},
  title        = {{Comparing different nonlinearities in readout systems for optical neuromorphic computing networks}},
  url          = {{http://doi.org/10.1038/s41598-021-03594-0}},
  volume       = {{11}},
  year         = {{2021}},
}
Altmetric
View in Altmetric
Web of Science
Times cited: