检索结果-内蒙古大学图书馆

Detection and recognition of images under interference conditions

JOURNAL OF optical TECHNOLOGY 2024年第8期91卷 547-555页

作者： Malakhova, Ekaterina yu. Shelepin, Konstantin yu. Shelepin, Yuriy E. Russian Acad Sci Pavlov Inst Physiol St Petersburg Russia

Subject of study. This study investigates human visual perception under noise conditions and models these mechanisms using generative neural networks. Aim of study. We aimed to understand how humans recognize visual signals under interference conditions and internal noise limitations and to develop effective computing technologies that implement heuristic principles of the visual system. Method. We analyzed psychophysical and neurophysiological data to understand perception efficiency, internal noise, and image reconstruction. We then modeled these perception mechanisms using generative neural networks. Main results. Our study examined visual perception thresholds in the presence of noise and characterized the properties of equivalent internal noise. We found that both internal noise and external uncertainty limit visual perception efficiency. Decreased efficiency serves as an indicator of information-processing disorders in sensory and cognitive pathology. Our work demonstrates that generative neural network models can effectively model key perceptual phenomena in noisy environments. Practical significance. Our computational models advance the understanding of noise-resistant perception mechanisms and facilitate differential diagnosis of visual processing disorders. (c) 2025 Optica Publishing Group

关键词： Image reconstruction Machine vision neural networks Pattern recognition Physiology Visual perception

来源：评论

学校读者我要写书评

暂无评论

Reservoir computing for equalization in a self-coherent receiver scheme

引用

OPTICS EXPRESS 2024年第23期32卷 40326-40339页

作者： Zelaci, Aimen Masaad, Sarah Bienstman, Peter Ghent Univ imec IDLab Dept Informat Technol Ghent Belgium Ghent Univ Imec Dept Informat Technol Photon Res Grp Ghent Belgium

Short-reach optical networks, the backbone of data centers, face a significant challenge: transmitting high data rates at low cost and low energy consumption. While coherent signals can carry high data rates, coherent receivers are expensive and complex. Also, to equalize channel dispersion, they rely on digital signal processing modules, which consume large amounts of power and introduce more latency. Photonic reservoirs emerged as a way to process these signals in the analog optical domain, alleviating the power consumption and latency issues in state-of-the-art receivers. In this work, we show in simulations that a photonic reservoir combined with a self-coherent photonic receiver achieves a BER of 3.8 x 10-3 for a 32 Gbaud 16-QAM signal and an 80 km link, requiring a low CSPR of 3 dB compared to state-of-the-art self-coherent receivers.

关键词： Fiber optic communications neural networks Nonlinear impairments optical networks optical signals Signal processing

来源：评论

学校读者我要写书评

暂无评论

Minimalist optical neural computing: optical Diffractive neural Network by 2-level Quantized Pixel-Wise optical Encoding

引用

LASER & PHOTONICS REVIEWS 2025年

作者： Liu, Xianjin Ma, Ting Bao, Qiwen Ma, Zhanying Gao, Guodong Xiao, Jun-Jun Harbin Inst Technol Shenzhen Coll Integrated Circuits Shenzhen 518055 Peoples R China Harbin Inst Technol Shenzhen Shenzhen Engn Lab Aerosp Detect & Imaging Shenzhen 518055 Peoples R China Harbin Inst Technol Shenzhen Educ Ctr Expt & Innovat Shenzhen 518055 Peoples R China

Diffractive optical neural networks (DONNs) offer high-speed, energy-efficient artificial intelligence (AI) computation but face challenges with optical misalignment and model-to-reality gaps. In this work, an ultra-simplified DONN architecture based on a digital mirror device (DMD) and camera, dubbed as m-DONN, is introduced and experimentally validated. Notably, within the m-DONN framework, the DMD acts as both the input layer and the solitary hidden layer, which is trained with 2-level quantization, markedly differing from the configuration found in traditional DONNs. This minimalism and binarization of the diffraction layer can result in a highly nonlinear correlation between the encoded input information and the output. A 10-classification accuracy of over 82% is achieved on the MNIST dataset in both theoretical modeling and experimental measurements, utilizing over 10 000 test samples. Furthermore, this m-DONN is employed to construct an online reinforcement learning agent capable of dynamically stabilizing a virtual inverted pendulum. The inherent simplicity of the proposed optical computing system, coupled with the cost-effective implementation using either active or passive key optical components, not only demonstrates an extremely powerful yet simple optical neuromorphic setup but also paves the way for the acceleration of optoelectronic AI applications across a variety of scenarios.

关键词： 2-level quantization digital mirror device dynamic control minimalist diffractive optical neural networks

来源：评论

学校读者我要写书评

暂无评论

Glaucoma Identification Using Convolutional neural networks Ensemble for Optic Disc and Cup Segmentation

引用

IEEE ACCESS 2024年 12卷 82720-82729页

作者： Virbukaite, Sandra Bernataviciene, Jolita Imbrasiene, Daiva Vilnius Univ Inst Data Sci & Digital Technol Vilnius 03101 Lithuania Lithuanian Univ Hlth Sci Inst Physiol & Pharmacol LT-44307 Kaunas Lithuania

Glaucoma is one of the diseases that can cause incurable blindness. The first noticeable symptoms appear only when the disease has progressed but an early diagnosis of the disease prevents the severe consequences of disease progression. In this paper, we developed a Convolutional neural networks (CNNs) based ensemble for joint optic disc (OD) and optic cup (OC) segmentation using modified Attention U-Net architecture with pre-trained ResNet34, ResNet50, MobileNet, Inceptionv3, DenseNet121 as backbones. The ensemble was trained on a mixed dataset consisting of REFUGE, Drishti-GS, and RIM-ONE r3 (RIM-ONE) datasets of eye fundus images and tested on images of each dataset separately. The most accurate joint OD and OC segmentation is achieved using an ensemble consisting of modified Attention U-Net with pre-trained ResNet34, Inceptionv3, and DenseNet121 as backbones and majority voting for final prediction. The highest Dice of 0.961 for OD and 0.894 for OC is achieved on the REFUGE test dataset, 0.974 for OD and 0.916 for OC on the Drishti-GS test dataset, and 0.978 for OD and 0.902 for OC on the RIM-ONE test dataset. The highest Intersection over Union of 0.925 for OD and 0.808 for OC is achieved on the REFUGE test dataset, 0.950 for OD and 0.845 for OC on the Drishti-GS test dataset, and 0.957 for OD and 0.822 for OC on the RIM-ONE test dataset. Using the segmentation results, the cup-to-disc ratio (CDR) has been calculated to classify eye fundus images into mild-stage, moderate-stage, severe-stage glaucoma, and non-glaucoma cases.

关键词： Glaucoma optical imaging optical computing optical variables measurement Convolutional neural networks Visualization Machine learning Ensemble networks glaucoma machine learning (ML) segmentation voting methods

来源：评论

学校读者我要写书评

暂无评论

Conflict resolution in nested transaction using fiber delay lines and load balancing with deep neural networks in optical data centers

引用

JOURNAL OF OPTICS-INDIA 2025年 1-16页

作者： Meenu Shanker, Udai Madan Mohan Malaviya Univ Technol Dept Comp Sci & Engn Gorakhpur Uttar Pradesh India

The exponential growth of data centers has led to increased challenges in transaction processing, particularly in the context of nested transactions where resource contention, deadlocks, and race conditions can cause significant performance degradation. These issues are especially pronounced in optical data centers, where the complexity of transaction management and the rapid pace of data transmission can exacerbate existing challenges. This paper presents a novel approach that integrates Fiber Delay Lines (FDLs) for effective conflict resolution with Deep neural networks (DNNs) for dynamic load balancing, specifically designed for optical data center environments. We present the mathematical foundations behind the integration of these technologies, highlighting how FDLs can help mitigate nested transaction conflicts by introducing controlled delays to resolve resource contention. Simultaneously, DNNs are employed to predict and balance the load dynamically, optimizing resource allocation and ensuring that the system operates at peak efficiency. The proposed approach not only addresses transaction conflicts but also adapts in real-time to fluctuating workloads, ultimately enhancing performance and throughput across the data center. This work presents both theoretical analysis and simulation implementation strategies for combining FDLs and DNNs, demonstrating their potential to transform resource management in high-demand, high-performance computing environments. Considering the proposed mechanism, the load balancing factor exceeds 0.2 even at a higher load of 0.85. At a load of 0.8, the transaction loss probability is approximately 10-3, with a delay of nearly 1.6 slots.

关键词： Conflict resolution Nested transactions Fiber delay lines (FDLs) Load balancing and deep neural networks (DNNs)

来源：评论

学校读者我要写书评

暂无评论

Demonstration of quantum dot reservoir computing based on spatio-temporal optical processing

引用

APPLIED OPTICS 2024年第28期63卷 G30-G36页

作者： Tate, Naoya Yamaguchi, Seiya Sakai, Shunichi Shimomura, Suguru Nishimura, Takahiro Kozuka, Jun Ogura, Yusuke Tanida, Jun Kyushu Univ Grad Sch Informat Sci & Elect Engn 744 MotookaNishi Ku Fukuoka 8190395 Japan Osaka Univ Grad Sch Informat Sci & Technol 1-5 Yamadaoka Suita Osaka 5650871 Japan Osaka Univ Grad Sch Engn 2-1 Yamadaoka Suita Osaka 5650871 Japan

A quantum dot network, which consists of coupled structures of randomly dispersed quantum dots, has been studied as a nano-scale optical reservoir for effective machine learning processing. In this study, we defined spatiotemporal fluorescence of a quantum dot network as the reservoir output, which is due to the characteristic dynamics of the excited energy in the network induced by laser pulse irradiation. In order to verify whether a quantum dot reservoir can improve the processing efficiency of advanced machine learning applications, we performed experimental reservoir computing using a numerical model. Several parameters that were required for the construction of the model were defined from the spatio-temporal fluorescence of an experimental quantum dot reservoir. Subsequently, the corresponding reservoir computing based on the model was numerically demonstrated. Reliable performances were successfully demonstrated as sufficient error rates toward the delayed XOR task. Additionally, the dependency on quantum dot compositions of these performances was clarified. (c) 2024 Optica Publishing Group

关键词： Fourier optics and optical processing Machine learning neural networks optical computing Photon counting Quantum dots

来源：评论

学校读者我要写书评

暂无评论

STADIA: Photonic Stochastic Gradient Descent for neural Network Accelerators

引用

ACM TRANSACTIONS ON EMBEDDED computing SYSTEMS 2023年第5期22卷 1-23页

作者： Xia, Chengpeng Chen, Yawen Zhang, Haibo Wu, Jigang Univ Otago Otago 9016 New Zealand Guangdong Univ Technol Guangzhou 510006 Guangdong Peoples R China

Deep neural networks (DNNs) have demonstrated great success in many fields such as image recognition and text analysis. However, the ever-increasing sizes of both DNN models and training datasets make deep leaning extremely computation- and memory-intensive. Recently, photonic computing has emerged as a promising technology for accelerating DNNs. While the design of photonic accelerators for DNN inference and forward propagation of DNN training has been widely investigated, the architectural acceleration for equally important backpropagation of DNN training has not been well studied. In this paper, we propose a novel silicon photonic-based backpropagation accelerator for high performance DNN training. Specifically, a general-purpose photonic gradient descent unit named STADIA is designed to implement the multiplication, accumulation, and subtraction operations required for computing gradients using mature optical devices including Mach-Zehnder Interferometer (MZI) and Mircoring Resonator (MRR), which can significantly reduce the training latency and improve the energy efficiency of backpropagation. To demonstrate efficient parallel computing, we propose a STADIA-based backpropagation acceleration architecture and design a dataflow by using wavelength-division multiplexing (WDM). We analyze the precision of STADIA by quantifying the precision limitations imposed by losses and noises. Furthermore, we evaluate STADIA with different element sizes by analyzing the power, area and time delay for photonic accelerators based on DNN models such as AlexNet, VGG19 and ResNet. Simulation results show that the proposed architecture STADIA can achieve significant improvement by 9.7x in time efficiency and 147.2x in energy efficiency, compared with the most advanced optical-memristor based backpropagation accelerator.

关键词： Stochastic gradient descent neural networks accelerator optical computing

来源：评论

学校读者我要写书评

暂无评论

Mobility-aware optical random waypoint and transfer learning-based load balancing

引用

INTERNATIONAL JOURNAL OF AD HOC AND UBIQUITOUS computing 2025年第2期48卷 94-109页

作者： Ramakrishnan, Arunkumar Balaiah, Thanasekhar Rajalakshmi Inst Technol Dept Comp Sci & Engn Chennai Tamil Nadu India Anna Univ Dept Comp Technol Chennai 600044 Tamil Nadu India

In recent years, a hopeful model of hybrid networks based on light fidelity (LiFi) and wireless fidelity (Wi-Fi) named hybrid LiFi-Wi-Fi networks (HLWNets) has been introduced. To address this issue, an innovative approach sewing training inspired optimisation_transfer learning (STIO_TL) is introduced for AP selection in the handover process. Initially, a system model of HLWNets is developed and the mobility-aware optical RWP is designed for the handover process. In the handover process, location is predicted every time using the deep recurrent neural network (DRNN). Afterwards, the AP selection is done by the proposed STIO_TL and is processed by several parameters. The proposed STIO_TL is the integration of the sewing circle inspired optimisation algorithm (STBO) and training-based optimisation algorithm (CIOA). Additionally, the effectiveness of the proposed STIO_TL is evaluated based on the evaluation metrics, like delay, handover occurrence, energy efficiency, and network throughput of 0.111 mS, 6.086, 0.099 Mbits/joules and 0.913 Mbps respectively.

关键词： sewing training inspired optimisation deep recurrent neural network DRNN transfer learning circle inspired optimisation algorithm multipath transmission control protocol

来源：评论

学校读者我要写书评

暂无评论

Time delay reservoir computing with a silicon microring resonator and a fiber-based optical feedback loop

引用

OPTICS EXPRESS 2024年第8期32卷 13419-13437页

作者： Donati, Giovanni Argyris, Apostolos Mancinelli, Mattia Mirasso, Claudio R. Pavesi, Lorenzo IFISC CSIC UIB Inst Fis Interdisciplinar & Sistemas Complejos Campus UIB Palma De Mallorca 07122 Spain Univ Trento Dept Phys Nanosci Lab Via Sommar 14 I-38123 Trento Italy

Silicon microring resonators serve as critical components in integrated photonic neural network implementations, owing to their compact footprint, compatibility with CMOS technology, and passive nonlinear dynamics. Recent advancements have leveraged their filtering properties as weighting functions, and their nonlinear dynamics as activation functions with spiking capabilities. In this work, we investigate experimentally the linear and nonlinear dynamics of microring resonators for time delay reservoir computing, by introducing an external optical feedback loop. After effectively mitigating the impact of environmental noise on the fiber -based feedback phase dependencies, we evaluate the computational capacity of this system by assessing its performance across various benchmark tasks at a bit rate of few Mbps. We show that the additional memory provided by the optical feedback is necessary to achieve error -free operation in delayed-boolean tasks that require up to 3 bits of memory. In this case the microring was operated in the linear regime and the photodetection was the nonlinear activation function. We also show that the Santa Fe and Mackey Glass prediction tasks are solved when the microring nonlinearities are activated. Notably, our study reveals competitive outcomes even when employing only 7 virtual nodes within our photonic reservoir. Our findings illustrate the silicon microring's versatile performance in the presence of optical feedback, highlighting its ability to be tailored for various computing applications. (c) 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement

关键词： CMOS technology Mach Zehnder interferometers neural networks optical resonators Ring lasers Semiconductor lasers

来源：评论

学校读者我要写书评

暂无评论

Seven Bit Nonvolatile Electrically Programmable Photonics Based on Phase-Change Materials for Image Recognition

引用

ACS PHOTONICS 2024年第2期11卷 723-730页

作者： Xia, Jian Wang, Tianci Wang, Zixuan Gong, Junjie Dong, Yunxiao Yang, Rui Miao, Xiangshui Huazhong Univ Sci & Technol Sch Integrated Circuits Wuhan 430074 Peoples R China Hubei Yangtze Memory Labs Wuhan 430205 Peoples R China

With the rapid development of the Internet of Things, how to efficiently store, transmit, and process massive amounts of data has become a major challenge now. optical neural networks based on nonvolatile phase change materials (PCMs) have become a breakthrough point due to their zero static power consumption, low thermal crosstalk, large-scale, and high efficiency. However, current photonic devices cannot meet the multilevel requirements in neuromorphic computing due to their limited storage capacity. Here, a new way for increasing storage capacity is paved from the perspective of modulation of the crystallization kinetics of PCMs. A more progressive transition from the amorphous to the crystalline states is achieved through the grain-refinement phenomenon induced by nitrogen (N) element doping in Ge2Sb2Te5 (GST), giving precise access to more multibit states. By integrating N-doped Ge2Sb2Te5 (N-GST) with a waveguide, a high-capacity nonvolatile photonic device enabling over 7 bits (similar to 222 levels) storage is achieved for the first time. The storage capacity is increased nearly by 7 times compared to the state-of-the-art device (similar to 32 levels). An optical convolutional neural network is successfully established for the MINIST handwritten digit recognition task by mapping synapse weight to the multiple optical levels, and a recognition accuracy of up to 96.5% is achieved. Our work provides a new strategy for the development of integrated photonic devices with multilevel and demonstrates enormous application potential in the field of large-scale photonic neural networks.

关键词： phase change materials photonic device multilevel optical neural networks storage capacity

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：