Scalable Optical Metasurface for on-Edge Visual Intelligence
We present a large-scale optical neural network (ONN) based on metasurface with 41 million photonic neurons, surpassing digital electronics in speed and efficiency. Our metasurface-based ONN matches ResNet and Vision Transformer performance, with <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$1,000 \times$</tex> energy-efficient processing compared to GPUs.
Authors
- Chaoran Huang (ORCID: https://orcid.org/0000-0001-6997-758X)
- Jiayong Peng (ORCID: https://orcid.org/0000-0002-7812-0585)
- Mingcheng Luo (ORCID: https://orcid.org/0009-0002-1484-213X)
- Chester Shu (ORCID: https://orcid.org/0000-0003-0020-6431)
Institutions
- Chinese University of Hong Kong (HK)
Publication Details
- Published
- 2025-06-29
- DOI
- https://doi.org/10.23919/oecc/psc62146.2025.11109982
- Primary Topic
- Metamaterials and Metasurfaces Applications
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- Shun Hing Institute of Advanced Engineering