Training an artificial neural network with backpropagation algorithms requires an extensive computational process. Our recent work proposes to implement the backpropagation algorithm optically for in-situ training of both the linear and nonlinear diffractive optical neural networks which enables the acceleration of training speed and improvement on the energy efficiency on core computing modules. We numerically validated that the proposed in-situ optical learning architecture achieves comparable accuracy to the in-silico training with an electronic computer on the task of object classification and matrix-vector multiplication, which further allows adaptation to the system imperfections. Besides, the self-adaptive property of our approach facilitates the novel application of the network for all-optical imaging through scattering media. The proposed approach paves the way for the robust implementation of large-scale diffractive neural networks to perform distinctive tasks all-optically.
Modern computer vision tasks are achieved by first capturing and storing large-scale images and then performing the processing electronically, the paradigm of which has the fundamentally limited speed and power efficiency with the continuous increase of the data throughput and computational complexity. We propose to build the all-optical artificial intelligent for light-speed computing, which performs advanced computer vision tasks during the imaging so that the detector can directly measure the computed results. The proposed method uses light diffraction property to build the optical neural network, where the neuron function is achieved by tuning the optical diffraction with a nonlinear threshold. Since every target scene has different frequency components, the proposed diffractive neural network is trained to perform various filtering on different frequency components and achieves different transform functions for the target scenes. We demonstrate the proposed approach can be used for high-speed detecting and segmenting visual saliency objects of the microscopic samples and macroscopic scenes as well as performing the task of object classification. The low power consumption, light-speed processing, and high-throughput capability of the proposed approach can serve as significant support for high-performance computing and will find applications in self-driving automobile, video monitoring, and intelligent microscopy, etc.