Multi-wavelength diffractive neural network with the weighting method

Recently, the diffractive deep neural network (D²NN) has demonstrated the advantages to achieve large-scale computational tasks in terms of high speed, low power consumption, parallelism, and scalability. A typical D²NN with cascaded diffractive elements is designed for monochromatic illumination. Here, we propose a framework to achieve the multi-wavelength D²NN (MW-D²NN) based on the method of weight coefficients. In training, each wavelength is assigned a specific weighting and their output planes construct the wavelength weighting loss function. The trained MW-D²NN can implement the classification of images of handwritten digits at multi-wavelength incident beams. The designed 3-layers MW-D²NN achieves a simulation classification accuracy of 83.3%. We designed a 1-layer MW-D²NN. The simulation and experiment classification accuracy are 71.4% and 67.5% at RGB wavelengths. Furthermore, the proposed MW-D²NN can be extended to intelligent machine vision systems for multi-wavelength and incoherent illumination.