Photonic Neural Network and In-Situ Training in a Synthetic Frequency Dimension

We develop a scalable photonic neural network utilizing the discrete frequency degree of freedom of light with the ability to train itself based on an in-situ backpropagation method with minimal reliance on external computers.