Abstract

The diffractive deep neural network (${{\rm D}^2}{\rm NN}$) has demonstrated its importance in performing various all-optical machine learning tasks, e.g.,  classification, segmentation, etc. However, deeper ${{\rm D}^2}{\rm NNs}$ that provide higher inference complexity are more difficult to train due to the problem of gradient vanishing. We introduce the residual ${{\rm D}^2}{\rm NNs}$ (Res-${{\rm D}^2}{\rm NN}$), which enables us to train substantially deeper diffractive networks by constructing diffractive residual learning blocks to learn the residual mapping functions. Unlike the existing plain ${{\rm D}^2}{\rm NNs}$, Res-${{\rm D}^2}{\rm NNs}$ contribute to the design of a learnable light shortcut to directly connect the input and output between optical layers. Such a shortcut offers a direct path for gradient backpropagation in training, which is an effective way to alleviate the gradient vanishing issue on very deep diffractive neural networks. Experimental results on image classification and pixel super-resolution demonstrate the superiority of Res-${{\rm D}^2}{\rm NNs}$ over the existing plain ${{\rm D}^2}{\rm NN}$ architectures.

© 2020 Optical Society of America

Multiple-view D²NNs array: realizing robust 3D object recognition

Jiashuo Shi, Liang Zhou, Taige Liu, Chai Hu, Kewei Liu, Jun Luo, Haiwei Wang, Changsheng Xie, and Xinyu Zhang
Opt. Lett. 46(14) 3388-3391 (2021)

In situ optical backpropagation training of diffractive optical neural networks

Tiankuang Zhou, Lu Fang, Tao Yan, Jiamin Wu, Yipeng Li, Jingtao Fan, Huaqiang Wu, Xing Lin, and Qionghai Dai
Photon. Res. 8(6) 940-953 (2020)

Broad-spectrum diffractive network via ensemble learning

Jiashuo Shi, Yingshi Chen, and Xinyu Zhang
Opt. Lett. 47(3) 605-608 (2022)

Orthogonality of diffractive deep neural network

Shuiqin Zheng, Shixiang Xu, and Dianyuan Fan
Opt. Lett. 47(7) 1798-1801 (2022)

Optical random phase dropout in a diffractive deep neural network

Yong-Liang Xiao, Sikun Li, Guohai Situ, and Zhisheng You
Opt. Lett. 46(20) 5260-5263 (2021)