Training of Photonic Neural Networks through In Situ Backpropagation
Tyler W. Hughes, Momchil Minkov, Ian A. D. Williamson, Yu Shi, Shanhui Fan · Conference on Lasers and Electro-Optics · 2019
We provide a protocol for training photonic neural networks based on adjoint methods. The gradient of the network with respect to its tunable degrees of freedom is computed by physically backpropagating an optical error signal.