All-optical autoencoder machine learning framework using linear diffractive processors

Abstract: Diffractive deep neural network (D2NN), known for its high speed and strong parallelism, has been widely applied across various fields, including pattern recognition, image processing, and image transmission. However, existing network architectures primarily focus on data representation within the original domain, with limited exploration of the latent space, thereby restricting the information mining capabilities and multifunctional integration of D2NNs. Here, we propose an all-optical autoencoder (OAE) framework that linearly encodes the input wavefield into a prior shape distribution in the diffractive latent space (DLS) and decodes the encoded pattern back to the original wavefield. By leveraging the bidirectional multiplexing property of D2NN, the OAE models function as encoders in one direction of wave propagation and as decoders in the opposite direction. We further apply the models to three key areas: image denoising, noise-resistant reconfigurable image classification, and image generation. Proof-of-concept experiments have been conducted to validate numerical simulations. Our OAE framework fully exploits the potential of latent representations, enabling a single set of diffractive processors to simultaneously achieve image reconstruction, representation, and generation. This work not only offers fresh insights into the design of optical generative models but also paves the way for developing multifunctional, highly integrated, and general optical intelligent systems.