Photonic Kernel Machine Learning for Ultrafast Spectral Analysis

We introduce photonic kernel machines, a scheme for ultrafast spectral analysis of noisy radio-frequency signals from single-shot optical intensity measurements. The approach combines the versatility of machine learning and the speed of photonic hardware to reach unprecedented throughput rates. We theoretically describe some of the key underlying principles, and then numerically illustrate the performance achieved in a photonic lattice-based implementation. We apply the technique both to picosecond pulsed radio-frequency signals, on energy-spectral-density estimation and a shape-classification task, and to continuous signals, on a frequency-tracking task. The optical computing scheme presented is resilient to noise while requiring minimal control on the photonic-lattice parameters, making it readily implementable in realistic state-of-the-art photonic platforms.