Realtime Inferences Every 125 ps Using a Semiconductor Laser Recurrent Neural Network
We experimentally implement a recurrent neural network (RNN) in an multimode semiconductor laser comprising 300 neurons and in-situ trainable weights. We demonstrate realtime inference at 8 Giga inferences/second and show realtime channel equalization at telecommunication wavelengths.
Authors
- Daniel Brunner (ORCID: https://orcid.org/0000-0002-4003-3056)
- Anas Skalli (ORCID: https://orcid.org/0000-0002-7039-926X)
- Romain Lance
Institutions
- Centre National de la Recherche Scientifique (FR)
- Franche-Comté Électronique Mécanique Thermique et Optique - Sciences et Technologies (FR)
Publication Details
- Published
- 2025-06-23
- DOI
- https://doi.org/10.1109/cleo/europe-eqec65582.2025.11110811
- Primary Topic
- Neural Networks and Reservoir Computing
- Type
- article
- Field-Weighted Citation Impact
- 0.00