Abstract
The increased demands of Deep Learning (DL) stress electronic computing hardware, prompting researchers in new computing paradigms. Neuromorphic photonic, emerged as a candidate, offering high throughput and energy efficiency, by harnessing light’s advantages. Here, we propose and experimentally demonstrate a Microdisk laser as a programmable all-optical activation function (AF) unit for photonic neural networks (PNNs). The device with a footprint of only 44.2 μm2 produced three non-linear AFs at 2Gbaud i.e., Inverse-ELU, Sigmoid and Clipped-GeLU at input power envelopes as low as 11.5 μW, exhibiting energy efficiency of 1.89 pJ/Symbol.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Shastri, B.J., et al.: Photonics for artificial intelligence and neuromorphic computing. Nat. Photon. 15 (2021)
Giamougiannis, G., et al.: Analog nanophotonic computing going practical: silicon photonic deep learning engines for tiled optical matrix multiplication with dynamic precision. Nanophotonics (2023)
Shen, Y., et al.: Deep learning with coherent nanophotonic circuits. Nat. Photon. 11(7), 441–446 (2017)
Nahmias, M.A., et al.: Photonic multiply-accumulate operations for neural networks. IEEE JSTQE 26(1), 1–18 (2020)
Totovic, A., et al.: WDM equipped universal linear optics for programmable neuromorphic photonic processors. NCE (2022)
Williamson, I.A.D., et al.: Reprogrammable electro-optic nonlinear activation functions for optical neural networks. IEEE JSTQE (2020)
Pappas, C., et al.: Programmable Tanh-, ELU-, Sigmoid-, and Sin-based nonlinear activation functions for neuromorphic photonics. IEEE J. Sel. Top. Quant. Electron. 29(6), 1–10 (2023). Photon. Sig. Process.
Kovaios, S., et al.: Programmable Tanh- and ELU-based photonic neurons in optics-informed neural networks. J. Lightwave Technol. (2024) https://doi.org/10.1109/JLT.2024.3366711
Mourgias-Alexandris, G., et al.: An all-optical neuron with sigmoid activation function. Opt. Express 27 (2019)
Shi, B., et al.: InP photonic integrated multi-layer neural networks: architecture and performance analysis. APL Photon. 7(1) (2022)
Jha, A., et al.: Programmable, high-speed all-optical nonlinear activation functions for neuromorphic photonics. OFC (2021)
Gordon, H.Y. L., et al.: All-optical ultrafast ReLU function for energy-efficient nanophotonic deep learning. Nanophotonics (2022)
Wu, B., et al.: Low-threshold all-optical nonlinear activation function based on a Ge/Si hybrid structure in a microring resonator. OMex (2022)
Demarchi, A., et al.: Low threshold hybrid InP-on-SOI microdisk laser diode with thin PIN junction. CSW (2023)
Acknowledgements
This work was supported by the EC via H2020 Projects PLASMONIAC (871391), SIPHO-G (101017194), ALLEGRO (101092766), and H2020 ERC HYPNOTIC (726420).
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2024 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Pappas, C. et al. (2024). An Ultra-Small InP Microdisk Laser Diode for Programmable Non-linear Activation Functions in Neuromorphic Photonics. In: Witzens, J., Poon, J., Zimmermann, L., Freude, W. (eds) The 25th European Conference on Integrated Optics. ECIO 2024. Springer Proceedings in Physics, vol 402. Springer, Cham. https://doi.org/10.1007/978-3-031-63378-2_66
Download citation
DOI: https://doi.org/10.1007/978-3-031-63378-2_66
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-63377-5
Online ISBN: 978-3-031-63378-2
eBook Packages: Physics and AstronomyPhysics and Astronomy (R0)