A Highly-Scalable Integrated Silicon Photonics Neural Network Architecture
A highly scalable optical neural network architecture based on the PQ Compress method is proposed. By using reported photonic components on a heterogeneous silicon pho-tonic platform, a minimal 332× chip size reduction is achieved when it scales up to a 784 × 784 network (MNIST).
Authors
- Liwei Duan (ORCID: https://orcid.org/0000-0002-5469-3741)
- Di Liang (ORCID: https://orcid.org/0000-0003-1366-0115)
- Yang Lan
Institutions
- University of Michigan (US)
Publication Details
- Published
- 2025-04-14
- DOI
- https://doi.org/10.1109/siphotonics64386.2025.10984695
- Primary Topic
- Photonic and Optical Devices
- Type
- article
- Field-Weighted Citation Impact
- 0.00