A multichannel optical computing architecture for advanced machine vision

作     者：ZHIHAO XU XIAOYUN YUAN TIANKUANG ZHOU LU FANG Zhihao Xu;Xiaoyun Yuan;Tiankuang Zhou;Lu Fang

作者机构：Sigma LaboratoryDepartment of Electronic EngineeringTsinghua UniversityBeijingChina Beijing National Research Center for Information Science and Technology(BNRist)BeijingChina Tsinghua Shenzhen International Graduate SchoolShenzhenChina Institute for Brain and Cognitive ScienceTsinqhua University(THUIBCS)BeijingChina 

出 版 物：《Light(Science & Applications)》 (光（科学与应用)（英文版）)

年 卷 期：2022年第11卷第9期

页      面：2235-2247页

核心收录：

学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 070207[理学-光学] 07[理学] 081104[工学-模式识别与智能系统] 08[工学] 0835[工学-软件工程] 0803[工学-光学工程] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 0702[理学-物理学] 

基　　金：supported in part by Ministry of Science and Technology of China under contract Na.20212D0109901 in part by Natural Science Foundation of China(NSFO under contract No.62125106 61860206003 and 62088102 in part by Bejing National Research Center for Information Science and Technology(BNRist)under Grant No.BNR2020RC01002 in part by Young Elite Scientists Sponsorship Program by CAST No.2021QNRC001.in part by Shuimu TSinghua Scholar Program China Postdoctoral Science Foundation No.2022M711874.and Postdoctoral International Exchange Program No.YJ20210124 

主　　题：computing multichannel projection 

摘      要：Endowed with the superior computing speed and energy efficiency,optical neural networks(ONNs)have attracted ever-growing attention in recent *** optical computing architectures are mainly single-channel due to the lack of advanced optical connection and interaction operators,solving simple tasks such as hand-written digit classification,saliency detection,*** limited computing capacity and scalability of single-channel ONNs restrict the optical implementation of advanced machine ***,we develop Monet:a multichannel optical neural network architecture for a universal multiple-input multiple-channel optical computing based on a novel projection-interference-prediction framework where the inter-and intra-channel connections are mapped to optical interference and *** our Monet,optical interference patterns are generated by projecting and interfering the multichannel inputs in a shared *** patterns encoding the correspondences together with feature embeddings are iteratively produced through the projection-interference process to predict the final output *** the first time,Monet validates that multichannel processing properties can be optically implemented with high-efficiency,enabling real-world intelligent multichannel-processing tasks solved via optical computing,including 3D/motion *** experiments on different scenarios demonstrate the effectiveness of Monet in handling advanced machine vision tasks with comparative accuracy as the electronic counterparts yet achieving a ten-fold improvement in computing *** intelligent computing,the trends of dealing with real-world advanced tasks are *** the capacity and scalability limitations of single-channel ONN and further exploring the multichannel processing potential of wave optics,we anticipate that the proposed technique will accelerate the development of more powerful optical Al as critical support for modern advanced mach