Details
| Originalsprache | Englisch |
|---|---|
| Fachzeitschrift | Nanophotonics |
| Frühes Online-Datum | 14 Feb. 2025 |
| Publikationsstatus | Elektronisch veröffentlicht (E-Pub) - 14 Feb. 2025 |
Abstract
Optical neural networks have demonstrated their potential to overcome the computational bottleneck of modern digital electronics. However, their development towards high-performing computing alternatives is hindered by one of the optical neural networks’ key components: the activation function. Most of the reported activation functions rely on opto-electronic conversion, sacrificing the unique advantages of photonics, such as resource-efficient coherent and frequency-multiplexed information encoding. Here, we experimentally demonstrate a photonic nonlinear activation function based on stimulated Brillouin scattering. It is coherent and frequency selective and can be tuned all-optically to take LEAKYRELU, SIGMOID, and QUADRATIC shape. Our design compensates for the insertion loss automatically by providing net gain as high as 20 dB, paving the way for deep optical neural networks.
ASJC Scopus Sachgebiete
- Biochemie, Genetik und Molekularbiologie (insg.)
- Biotechnologie
- Werkstoffwissenschaften (insg.)
- Elektronische, optische und magnetische Materialien
- Physik und Astronomie (insg.)
- Atom- und Molekularphysik sowie Optik
- Ingenieurwesen (insg.)
- Elektrotechnik und Elektronik
Zitieren
- Standard
- Harvard
- Apa
- Vancouver
- BibTex
- RIS
in: Nanophotonics, 14.02.2025.
Publikation: Beitrag in Fachzeitschrift › Artikel › Forschung › Peer-Review
}
TY - JOUR
T1 - All-optical nonlinear activation function based on stimulated Brillouin scattering
AU - Slinkov, Grigorii
AU - Becker, Steven
AU - Englund, Dirk
AU - Stiller, Birgit
N1 - Publisher Copyright: © 2025 the author(s),
PY - 2025/2/14
Y1 - 2025/2/14
N2 - Optical neural networks have demonstrated their potential to overcome the computational bottleneck of modern digital electronics. However, their development towards high-performing computing alternatives is hindered by one of the optical neural networks’ key components: the activation function. Most of the reported activation functions rely on opto-electronic conversion, sacrificing the unique advantages of photonics, such as resource-efficient coherent and frequency-multiplexed information encoding. Here, we experimentally demonstrate a photonic nonlinear activation function based on stimulated Brillouin scattering. It is coherent and frequency selective and can be tuned all-optically to take LEAKYRELU, SIGMOID, and QUADRATIC shape. Our design compensates for the insertion loss automatically by providing net gain as high as 20 dB, paving the way for deep optical neural networks.
AB - Optical neural networks have demonstrated their potential to overcome the computational bottleneck of modern digital electronics. However, their development towards high-performing computing alternatives is hindered by one of the optical neural networks’ key components: the activation function. Most of the reported activation functions rely on opto-electronic conversion, sacrificing the unique advantages of photonics, such as resource-efficient coherent and frequency-multiplexed information encoding. Here, we experimentally demonstrate a photonic nonlinear activation function based on stimulated Brillouin scattering. It is coherent and frequency selective and can be tuned all-optically to take LEAKYRELU, SIGMOID, and QUADRATIC shape. Our design compensates for the insertion loss automatically by providing net gain as high as 20 dB, paving the way for deep optical neural networks.
KW - Brillouin scattering
KW - nonlinear activation function
KW - nonlinear optics
KW - optical fiber
KW - optical neural network
KW - optoacoustics
KW - photonic neuromorphic computing
UR - http://www.scopus.com/inward/record.url?scp=85218165005&partnerID=8YFLogxK
U2 - 10.1515/nanoph-2024-0513
DO - 10.1515/nanoph-2024-0513
M3 - Article
AN - SCOPUS:85218165005
JO - Nanophotonics
JF - Nanophotonics
SN - 2192-8606
ER -