From Event: SPIE LASE, 2023
We introduce an ultralow latency computing technique that utilizes femtosecond pulses for the classification of optical data. Spectral mapping of data onto femtosecond pulses and transformation utilizing the Nonlinear Schrodinger Kernel reduces the latency in data classification by several orders of magnitude and increases inference accuracy in experiments. Closed-loop optimization and training of the optical nonlinearities is achieved using spectral phase-encoding and leads to improvement in the accuracy of data classification in time-stretch microscopy.
© (2023) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Tingyi Zhou and Bahram Jalali, "Realtime computing with femtosecond laser pulses," Proc. SPIE PC12406, Real-time Measurements, Rogue Phenomena, and Single-Shot Applications VIII, PC1240604 (Presented at SPIE LASE: February 01, 2023; Published: 17 March 2023); https://doi.org/10.1117/12.2656434.6321500676112.