Optical Neural Networks with Tensor Compression and Photonic Memory

Xian Xiao, Stanley Cheung, Bassem Tossoun, Thomas Van Vaerenbergh, Geza Kurczveil, Raymond G. Beausoleil

Abstract: 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.