12‐2: Liquid Crystal Based All Optical Diffraction Neural Networks

In this paper, we propose the liquid crystal based all optical diffraction neural network. Based on the optical diffractive propagation theory, the photo‐induced aligned liquid crystal volume phase plates work as the neural layers while the single pixel of each liquid crystal layer denote the neural with modulating the geometric phase retardation for the incident light. The simulation and experimental results are conducted to demonstrate the neural network for the application of digital handwritten digits’ recognition, achieving 89% recognition accuracy of the handwritten digits through the all optical diffraction neural network, presenting the potential application in image classification, logical operations, and other fields.