Optical Neural Networks with Tensor Compression and Photonic Memory
Xian Xiao, Stanley S. Cheung, Bassem M. Tossoun, Thomas Van Vaerenbergh, Géza Kurczveil, Raymond G. Beausoleil · 2024
This paper introduces our recent efforts on scalable, energy-efficient, and low-latency tensorized optical neural networks, including design considerations, options for wavelength-parallel photonic tensor cores, and photonic memory for non-volatile tuning.