检索结果-内蒙古大学图书馆

Rapid configuring method for a programmable photonic integrated circuit based on a tandem neural network

OPTICS LETTERS 2025年第5期50卷 1731-1734页

作者： Fan, Zeyang Dan, Yihang Lin, Junmin Zhang, Tian Dai, Jian Xu, Kun Beijing Univ Posts & Telecommun State Key Lab Informat Photon & Opt Commun Beijing 100876 Peoples R China China Mobile Commun Grp Shaanxi Co Ltd Customer Response Ctr Xian 710077 Peoples R China China Telecom Unmanned Technol Jiangsu Co Ltd Prod & Res & Dev Dept Jiangsu 210019 Peoples R China

Programmable photonic integrated circuits (PPICs), as optical analog matrix multipliers, emerge as a leading candidate of a revolutionary technology. However, the efficient voltage configuration of programmable devices in the circuit presents a significant challenge to its development. Here, we propose a black-box method based on tandem neural network to rapidly predict the voltage configuration of arbitrary matrices. We experimentally demonstrate the feasibility of our method on a 4 x 4 PPIC, achieving the average fidelity of 0.989 for 10,000 matrices. Furthermore, we experimentally implement an optical-electric hybrid model based on our method, obtaining a training accuracy of 97.59% on including for text and data mining (TDM), Artificial Intelligence (AI) training, and similar technologies, are reserved.

关键词： Deep learning neural networks optical computing optical neural systems optical signals Phase shift

来源：评论

学校读者我要写书评

暂无评论

Efficient Design Optimization for Diffractive Deep neural networks

引用

IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS 2025年第3期44卷 1199-1203页

作者： Wu, Kun Liu, Yuncheng Gao, Hui Tao, Jun Xiong, Wei Li, Xin Wuhan Univ Sch Elect Informat Wuhan 430072 Peoples R China Huazhong Univ Sci & Technol Natl Lab Optoelect Wuhan 430074 Peoples R China Huazhong Univ Sci & Technol Sch Optic & Elect Informat Wuhan 430074 Peoples R China Optic Valley Lab Hubei Wuhan 430074 Peoples R China Fudan Univ Sch Microelect State Key Lab Integrated Chips & Syst Shanghai 200433 Peoples R China Duke Kunshan Univ Data Sci Res Ctr Kunshan 215316 Jiangsu Peoples R China

Since diffractive deep neural network (D2NN) provides a full optical solution to implement deep neural networks (DNNs), it offers ultrafast operation speed and virtually unlimited bandwidth, yielding an alternative-yet-competitive approach for computer-based neural networks. A D2NN is composed of several 3D-printed phase masks as hidden layers and a number of optical detectors at the output. To enable automatic and efficient design of D2NNs, we propose an iterative optimization method to determine the optimal design parameters of D2NNs. During each iteration step, we first optimize the physical parameters for masks (e.g., thicknesses) while fixing the detector parameters (e.g., locations). Next, we exhaustively search the detector parameters with fixed masks. These two steps are repeated until convergence is reached. Our numerical experiments demonstrate that the proposed optimization algorithm can produce a high-performance D2NN achieving 97% accuracy for recognizing handwritten digits.

关键词： Detectors Artificial neural networks optical detectors optical diffraction Ultrafast optics optical interferometry Integrated optics Integrated circuits optical resonators optical computing Diffractive deep neural network ((DNN)-N-2) optical neural networks (ONNs) optimization

来源：评论

学校读者我要写书评

暂无评论

Real-time target recognition with all-optical neural networks for ghost imaging

引用

OPTICS EXPRESS 2024年第23期32卷 40967-40978页

作者： Xi, Yuanyuan He, Yuchen Wang, Yadi Chen, Hui Zheng, Huaibin Liu, Jianbin Zhou, Yu Xu, Zhuo Xi An Jiao Tong Univ Inst Artificial Intelligence & Robot Xian 710049 Peoples R China Xi An Jiao Tong Univ Elect Mat Res Lab Key Lab Sch Elect Sci & EngnMinist Educ Xian 710049 Peoples R China Xi An Jiao Tong Univ Int Ctr Dielect Res Sch Elect Sci & Engn Xian 710049 Peoples R China Xi An Jiao Tong Univ Dept Appl Phys MOE Key Lab Nonequilibrium Synth & Modulat Condens Xian 710049 Peoples R China

The generation and structural characteristics of random speckle patterns impact the implementation and imaging quality of computational ghost imaging. Their modulation is limited by traditional electronic hardware. We aim to address this limitation using the features of an all-optical neural network. This work proposes a real-time target recognition system based on an all-optical diffraction deep neural network for ghost imaging. We use a trained neural network to perform pure phase modulation on visible light, and directly complete the target recognition task by detecting the maximum value of light intensity signals at different positions. We optimized the system by simulating the effects of parameters, such as the number of layers of the network, photosensitive pixel, unit area etc., on the final recognition performance, and the accuracy of target recognition reached 91.73%. The trained neural network is materialised by 3D printing technology and experiments confirmed that the system successfully performs real-time target recognition at a low sampling rate of 1.25%. It also verified the feasibility and noise resistance of the system in practical application scenarios.

关键词： Computational imaging Ghost imaging neural networks optical computing optical neural systems Spatial light modulators

来源：评论

学校读者我要写书评

暂无评论

Energy-efficient photonic neural networks for high-speed AI computation

引用

JOURNAL OF OPTICS-INDIA 2025年 1-9页

作者： Palai, Gopinath Kumar, Bhukya Arun Mishra, Bibhu Kalyan Laha, Jasobanta Satpathy, Rabinarayan Sri Sri Univ Fac Engn & Technol Cuttack 756004 Odisha India Lovely Profess Univ Phagwara Punjab India

Photonic neural networks (PNNs) are an innovative computational paradigm that leverages the speed and efficiency of light to perform neural network operations. By utilizing components like lasers, Mach-Zehnder Interferometers (MZIs), and optical waveguides, PNNs can achieve computation speeds in the terahertz (THz) range, far exceeding the gigahertz (GHz) speeds of traditional electronic neural networks. This speed advantage is coupled with exceptional energy efficiency, as PNNs eliminate energy-intensive electrical-to-optical conversions and reduce resistive losses, making them ideal for large-scale applications such as data *** a 4-neuron PNN, for instance, an input power vector (1, 0.8, 0.6, 0.4) represents varying light intensities fed into the system. The weights matrix, determined randomly, influences phase shifts in the MZIs, ranging from 0 to pi radians. The output power is governed by the cosine-squared function of these phase shifts, leading to dynamic variations in the output. For example, neuron 1's output power decreases from 1.0 to 0.5, while neuron 3 experiences significant amplification, with output power rising from 0.6 to 2.2. Each neuron consumes only 0.1 picojoules (pJ), resulting in a total energy consumption of 0.4 pJ for the entire network. These results demonstrate PNNs' ability to process optical signals with remarkable efficiency, offering both lower energy consumption and faster speeds compared to electronic neural networks. This makes PNNs a promising solution for developing energy-efficient, high-speed AI hardware, capable of meeting the growing demands of modern computational tasks. By combining the advantages of light-based computation with the scalability of neural networks, PNNs pave the way for transformative advancements in artificial intelligence and machine learning, enabling faster and more sustainable computing solutions for the future.

关键词： Photonic neural networks Energy Mach-Zehnder interferometer

来源：评论

学校读者我要写书评

暂无评论

Deep photonic reservoir computing recurrent network

引用

OPTICA 2023年第12期10卷 1745-1751页

作者： Shen, Yi-Wei Li, Rui-Qian Liu, Guan-Ting Yu, Jingyi He, Xuming Yi, Lilin Wang, Cheng ShanghaiTech Univ Sch Informat Sci & Technol Shanghai 201210 Peoples R China Shanghai Jiao Tong Univ Sch Elect Informat & Elect Engn Dept Elect Engn State Key Lab Adv Opt Commun Syst & Networks Shanghai 200240 Peoples R China ShanghaiTech Univ Shanghai Engn Res Ctr Energy Efficient & Custom A Shanghai 201210 Peoples R China

Deep neural networks usually process information through multiple hidden layers. However, most hardware reservoir computing recurrent networks only have one hidden reservoir layer, which significantly limits the capability of solving practical complex tasks. Here we show a deep photonic reservoir computing (PRC) architecture, which is constructed by cascading injection-locked semiconductor lasers. In particular, the connection between successive hidden layers is all optical, without any optical-electrical conversion or analog-digital conversion. The proof of concept PRC consisting of 4 hidden layers and a total of 320 interconnected neurons (80 neurons per layer) is demonstrated in experiment. The deep PRC is applied in solving the real-world problem of signal equalization in an optical fiber communication system. It is found that the deep PRCexhibits strong capability in compensating for the nonlinear impairment of optical fibers.

关键词： Fiber optic communications Information processing Nonlinear optical fibers optical computing optical neural systems Semiconductor lasers

来源：评论

学校读者我要写书评

暂无评论

Hybrid Deep Photonic Spiking neural Network for Automatic Modulation Recognition

引用

JOURNAL OF LIGHTWAVE TECHNOLOGY 2025年第6期43卷 2672-2680页

作者： Zhang, Yahui Huang, Zhiquan Xiang, Shuiying Guo, Xingxing Zhang, Wu Tan, Qinggui Han, Genquan Hao, Yue Xidian Univ State Key Lab Integrated Serv Networks Xidian 710071 Peoples R China Xidian Univ Sch Microelect State Key Discipline Lab Wide Bandgap Semicond Tec Xian 710071 Peoples R China China Acad Space Technol CAST Xian Xian 100081 Peoples R China

Spiking neural networks (SNNs) possess remarkable capabilities in processing spatial and temporal information. Photonic SNNs, combining the advantages of high-bandwidth and low-latency of photonics, are highly-efficient, high-speed neural processors. However, training a deep photonic SNN, or even a deep SNN, for many complex tasks is challenging. In this work, we propose a hybrid deep photonic SNN (HDPSNN) for the recognition of automatic modulation recognition (AMR). The proposed HDPSNN integrates existing convolutional and long short-term memory layers for feature extraction, SNNs for recognition, and photonic SNN for accelerating the recognition process. In the HDPSNN, a two-layers photonic SNN is demonstrated experimentally using vertical-cavity surface-emitting lasers neurons and Fabry-P & eacute;rot laser with an embedded saturable absorber neuron. For the hybrid deep SNN (HDSNN), numerical results show that the maximum accuracy reaches 89.59% with the signal-to-noise ratios (SNRs) of +18 dB. Additionally, the inference accuracy reaches 80% for 100 randomly selected samples at SNR of 0 dB. For the HDPSNN, experimental results indicate AMR accuracy of 70% for the inference of the same 100 randomly selected samples. Thus, the results shown that the proposed HDPSNN can achieve AMR with minor accuracy loss. These findings suggest a method to combine deep neural network, SNN and photonics to achieve complex tasks in optical SNN. It is significant to pursue ultra-low delay and solutions for complex tasks, design neural structures of photonic SNNs, and realize hardware-algorithm collaborative computing.

关键词： Neurons Photonics Vertical cavity surface emitting lasers Feature extraction optical attenuators Accuracy Training Modulation Spiking neural networks optical modulation Automatic modulation recognition photonic spiking neural networks VCSELs

来源：评论

学校读者我要写书评

暂无评论

Neuromorphic Network with Photonic Weighting and Photoelectronic Nonlinear Activation Based on SOA and APD

引用

ACS PHOTONICS 2024年第10期11卷 4193-4199页

作者： Zheng, Dianzhuang Xiang, Shuiying Li, Nianqiang Zhang, Yahui Guo, Xingxing Zhao, Liyan Hao, Yue Xidian Univ State Key Lab Integrated Serv Networks Xian 710071 Peoples R China Xidian Univ Sch Microelect State Key Discipline Lab Wide Band Gap Semicond Te Xian 710071 Peoples R China Soochow Univ Sch Optoelect Sci & Engn Suzhou 215000 Peoples R China

Photonic neuromorphic computing is emerging as a promising approach for low-latency, energy-efficient nonvon Neumann computing systems. Nonlinear activation functions are key components of photonic neural networks. However, the current implementation of nonlinear activation functions faces issues such as high power consumption, high threshold, and limited bandwidth. Implementing efficient nonlinear activation functions in photonic neural networks remains a significant challenge. Here, we propose and demonstrate for the first time a photonic-electronic neural network consisting of semiconductor optical amplifiers (SOAs) as photonic synapses and InGaAs avalanche photodiodes (APDs) with trans-impedance amplifiers as photoelectronic nonlinear neurons. The APD achieves a tanh-like nonlinear activation function while performing photoelectronic conversion without additional electronic processing. It has a large operating bandwidth and a response frequency of 8 GHz, with a threshold input optical power of -30 dBm. Utilizing the proposed linear weighting mechanism of SOAs, low-power input signals can be weighted while being provided with gain. Embedding the measured nonlinear activation function into convolutional neural networks with two convolutional layers results in inference accuracies of 98.485% and 91.2% for the classification tasks of the MNIST handwritten digit image data set and the Fashion-MNIST image data set, respectively. Furthermore, leveraging the linear weighting mechanism of SOAs and nonlinear activation of APDs, we have successfully accomplished hardware inference tasks on the proposed network architecture. In addition, the computational speed and energy efficiency of the proposed network architecture are evaluated. We believe this work holds promise for advancing the development of monolithically integrated large-scale photonic-electronic intelligent neuromorphic systems in the future.

关键词： neuromorphic computing photoelectronic nonlinearactivation semiconductor optical amplifiers photonic-electronicneural network the InGaAs avalanche photodiodes

来源：评论

学校读者我要写书评

暂无评论

LUMF-YOLO: a lightweight object detection network integrating UAV motion features

引用

computing 2025年第1期107卷 1-22页

作者： Wang, Sicheng Li, Gang He, Bin Cheng, Bin Ding, Yulong Li, Wei Tongji Univ Coll Elect & Informat Engn Shanghai 201804 Peoples R China Shanghai Res Inst Intelligent Autonomous Syst Shanghai 200120 Peoples R China Shanghai Sunshine Rehabil Ctr Shanghai 201613 Peoples R China Natl Key Lab Autonomous Intelligent Unmanned Syst Shanghai 200120 Peoples R China

In recent years, the network structure has become more complex to improve the accuracy of convolutional neural networks (CNN), increasing computing power requirements. However, the edge computing capability of unmanned aerial vehicles (UAVs) is low, which makes it challenging to meet the computing power requirements of large neural networks. Therefore, achieving real-time and accurate multi-object detection on UAVs with limited computing resources becomes more challenging. To solve this problem, we present LUMF-YOLO, a lightweight object detection network for UAVs with limited computing power. We employ MobileNetV2 as the backbone, replacing CSPDarknet53 to reduce parameter count and enhance inference speed. We integrate PWC-Net for optical flow estimation and fuse its features with the first, third, and sixth layers of the feature pyramid to improve small object detection. We remove the layer with the largest receptive field from YOLOv4 to focus on detecting smaller targets. An adaptive copy-paste method is introduced to expand the dataset for small and underrepresented objects. LUMF-YOLO achieves 43.3% mAP@0.5 on the VisDrone dataset, demonstrating superior accuracy and performance compared to existing lightweight networks.

关键词： Convolutional neural networks Lightweight Object detection UAV motion features fusion

来源：评论

学校读者我要写书评

暂无评论

Efficient on-chip training of large-scale optical neural network through block adjoint training algorithm

引用

OPTICS EXPRESS 2024年第26期32卷 46633-46648页

作者： Yang, Zhiwei Zhang, Tian Dai, Jian Xu, Kun Beijing Univ Posts & Telecommun State Key Lab Informat Photon & Opt Commun Beijing 100876 Peoples R China Beijing Univ Posts & Telecommun Sch Elect Engn Beijing 100876 Peoples R China

MZI-based block optical neural networks (BONNs), which utilize block matrix multiplication to achieve large-scale network models, have garnered significant attention but still lack efficient training algorithms. In this article, by calculating the original field and adjoint field for the block matrices in BONNs and directly updating the phase values of all phase shifters within the optical mesh, we propose an on-chip block adjoint training (BAT) algorithm for large-scale BONNs. To demonstrate the effectiveness of our proposed algorithm, the trained BONNs are applied in image classification tasks for MNIST and SVHN datasets. The calculated results demonstrate that the performance of the BAT algorithm (95.915% for the MNIST dataset and 82.64% for the SVHN dataset) is competitive with the traditional gradient algorithm based on artificial neural networks (96.238% and 84.182%), but the BONNs can infer 1.5 times and 1.3 times faster than artificial neural networks, respectively. By studying the influence of the block size and the inputted position of the padded zero signals, we demonstrate that the BAT algorithm based on the BONNs with 12 block sizes can achieve higher performance by adding the padded zero signals to the same side beside the normal inputted signals. Additionally, we demonstrate that substituting the complete weight matrices with unitary matrices to construct BONNs is an efficient way to reduce both the system area and the required trainable parameters. Finally, we demonstrate the relatively good robustness of the BAT algorithm and the imprecision alleviation method by using on-chip retraining. Notably, our proposed BAT algorithm shows excellent potential for more complex tasks and network models.

关键词： Genetic algorithms Machine vision neural networks optical neural systems Phase shift Stochastic gradient descent

来源：评论

学校读者我要写书评

暂无评论

Reliable adaptive edge-cloud collaborative DNN inference acceleration scheme combining computing and communication resources in optical networks

引用

JOURNAL OF optical COMMUNICATIONS AND NETWORKING 2023年第10期15卷 750-764页

作者： Yin, Shan Jiao, Yurong You, Chenyu Cai, Mengru Jin, Tianyu Huang, Shanguo Beijing Univ Posts & Telecommun State Key Lab Informat Photon & Opt Commun Beijing 100876 Peoples R China

With the continuous development of the Artificial Intelligence of Things, deep neural network (DNN) models require a larger amount of computing capacity. The emerging edge-cloud collaboration architecture in optical networks is proposed as an effective solution, which combines edge computing with cloud computing to provide faster response and reduce the cloud load for compute-intensive tasks. The multi-layered DNN model can be divided into subtasks that are offloaded to edge and cloud servers for computation in this architecture. In addition, as bearer networks for computing capacity, once a server or link in optical networks fails, a large amount of data can be lost, so the robust reliability of the edge-cloud collaborative optical networks is very important. To solve the above problems, we design a reliable adaptive edge-cloud collaborative DNN inference acceleration scheme (RACAI) combining computing and communication resources. We formulate the RACAI into a mixed integer linear programming model and develop a multi-agent deep reinforcement learning algorithm (MADRL-RACIA) to jointly optimize DNN task partitioning, offloading, and protection. The simulation results show that compared with the benchmark schemes, the proposed MADRL-RACIA can provide a guarantee of reliability for more tasks under latency constraints and reduce the blocking probability.

关键词： Task analysis Servers Collaboration optical fiber networks Computational modeling Reliability Cloud computing

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：