Diffractive optical neural networks for intelligent informatics: image processing, communications, and quantum applications
作者
Yuchen Ma,Zhuochao Wang,Jiancheng An,Xing Lin,Jingtian Hu,Can Huang,Chau Yuen,Liangcai Cao,Guangwei Hu
出处
期刊:Advanced photonics [SPIE - International Society for Optical Engineering] 日期:2026-09-12卷期号:8 (05): 1-20被引量:1
标识
DOI:10.1117/1.ap.8.5.054001
摘要
Diffractive optical neural networks (DONNs) represent a physical computing paradigm that performs machine learning tasks by programming the propagation of light through a series of engineered diffractive layers. By harnessing the principles of diffraction, DONNs execute complex computations through a single pass of light propagation with massive parallelism and minimal power consumption, offering a compelling alternative to conventional electronic architectures. In this review, we summarize the fundamental principles of DONNs and examine their applications across key domains, ranging from optical image classification and high-dimensional imaging to communication-capacity expansion through multiplexing and emerging uses in quantum information processing. Finally, we discuss the main challenges for practical implementation of DONNs and outline possible future directions for a new generation of intelligent systems that compute with light.