检索结果-内蒙古大学图书馆

Energy-Efficient Programable Analog Computing: Analog computing in a standard CMOS process

IEEE Solid-State Circuits Magazine 2024年第4期16卷 32-40页

作者： Hasler, Jennifer Georgia Institute of Technology School of Electrical and Computer Engineering AtlantaGA30332 United States

When I started working on analog computing for neural network systems in the 1980s, the question everyone feared to be asked at the end of their presentation was "couldn't this be done on a DSP processor?"Carver Mead provided a theoretical and intuitive answer in 1990 that analog computing should be thousands of times more efficient than digital computation. Three decades later the "couldn't this be done on a DSP?"question, or its related question "couldn't this be done on a digital computing IC?", has been definitively addressed with a resounding No for several implementations. This paper addresses how end-to-end analog computing system demonstrates this increased efficiency over digital computing systems. These computations are possible entirely in a standard CMOS process using Floating-Gate (FG) devices available in any CMOS IC process. The efficiency is both at the computational circuit level, as well as at the architecture level. The first crossbar and single-transistor synapse concept appeared 30 years ago. The definitive demonstration of Mead's energy efficiency hypothesis appeared 20 years ago, and before that demonstration was the start of Compute in Memory (CiM) as well as the start of a large-scale FPAA with FG-enabled connections. Analog computing opportunities in Standard CMOS IC processes can provide many opportunities for the wider commercial market. © 2009-2012 IEEE.

关键词： CMOS integrated circuits

来源：评论

学校读者我要写书评

暂无评论

Ensemble deep learning for high-precision classification of 90 rice seed varieties from hyperspectral images

引用

Journal of Ambient Intelligence and Humanized Computing 2024年第6期15卷 2883-2899页

作者： Taheri, AmirMasoud Ebrahimnezhad, Hossein Sedaaghi, Mohammadhossein Computer Vision Research Laboratory Faculty of Electrical Engineering Sahand University of Technology Tabriz Iran

To develop rice varieties with better nutritional qualities, it is important to classify rice seeds accurately. Hyperspectral imaging can be used to extract spectral information from rice seeds, which can then be used to classify them into different varieties. The challenges of precise classification increase when there are many classes and few training samples. In this paper, we present a novel method for high-precision Hyperspectral Image (HSI) classification of 90 different classes of rice seeds using ensemble deep learning. Our method first employs band selection techniques to select the optimal hyperspectral bands for rice seed classification. Then, a deep neural network is trained with the selected hyperspectral and RGB data from rice seed images to obtain different models for different bands. Finally, an ensemble of deep learning models is employed to classify rice seed images and improve classification accuracy. The proposed method achieves an overall precision ranging from 92.73 to 96.17% despite a large number of classes and low data samples for each class and with only 15 selected hyperspectral bands. This precision is significantly higher than the state-of-the-art classical machine learning methods like random forest, confirming the effectiveness of the proposed method in classifying hyperspectral images of rice seeds. © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2024.

关键词： Image classification

来源：评论

学校读者我要写书评

暂无评论

Super-resolution reconstruction of off-axis digital holography based on Real-ESRGAN

引用

Applied Optics 2025年第9期64卷 2245-2254页

作者： Chen, Geng Chen, Jie Liao, Houzhang Liu, Keke Tang, Xin Kong, Yong School of Electric and Electrical Engineering Shanghai University of Engineering Science Shanghai201620 China School of Optical-Electrical and Computer Engineering University of Shanghai for Science and Technology Shanghai200093 China

Off-axis digital holography plays a crucial role in high-precision three-dimensional imaging. However, high-resolution phase images are often affected by the limited pixel size of the sensor. To address this issue, this study proposes, for the first time, to the best of our knowledge, the application of Real-ESRGAN for the super-resolution reconstruction of off-axis digital holography images. By directly applying super-resolution processing to the hologram, the limitations of the sensor's pixel size are overcome, followed by phase reconstruction to obtain high-resolution phase images. The Real-ESRGAN model enhances the texture information of the hologram through a deep network with multiple residual-in-residual dense blocks (RRDBs), restoring a clearer hologram. A U-Net architecture with spectral normalization, offering stronger discriminative capability, is used to provide detailed per-pixel feedback for the generator. The experimental results demonstrate that this method can be applied to multi-scale holograms, outperforming comparison networks in both visual quality and quantitative metrics, thus providing an innovative solution for the super-resolution reconstruction of off-axis digital holography. © 2025 Optica Publishing Group. All rights, including for text and data mining (TDM), Artificial Intelligence (AI) training, and similar technologies, are reserved.

关键词： Holograms

来源：评论

学校读者我要写书评

暂无评论

Dual-Band Implantable Antenna Loaded with Patch Slots for Wireless Biotelemetry Systems

引用

Progress In Electromagnetics Research C 2024年 141卷 151-162页

作者： Rana, Md. Masud Islam, Md. Ariful Mehedi, Ibrahim M. Department of Electrical and Electronic Engineering Rajshahi University of Engineering and Technology Rajshahi6204 Bangladesh Department of Electrical and Computer Engineering King Abdulaziz University Jeddah Saudi Arabia

This paper presents a dual-band implantable antenna with coaxial probe feeding for wireless biotelemetry applications. The antenna features spiral patches, resulting in a compact size of 27 × 14 × 1.6 mm3. It can operate in two different frequency bands, 241–641 MHz and 1.17–2.06 GHz, providing coverage for the medical implant communication service (MICS) band and the industrial, scientific, and medical (ISM) band. This simple design offers improved return loss and higher bandwidths that are achieved by incorpo-rating patch slots and shortening pins in spiral patches, representing a significant contribution to the field of dual-band antenna design for wireless biotelemetry systems. In the human body phantom model, the SAR values of 48.9 mW/kg and 1.19 W/kg are achieved, which satisfy the IEEE standard safety constraints. An experimental prototype of the proposed antenna is fabricated, demonstrating acceptable return loss and VSWR. © 2024, Electromagnetics Academy. All rights reserved.

关键词： IEEE Standards

来源：评论

学校读者我要写书评

暂无评论

An End-to-End Two-Branch Network Towards Robust Video Fingerprinting

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2024年第5期5卷 2371-2384页

作者： Xu, Yingying Zhou, Yuanding Li, Xinran Zhao, Gejian Qin, Chuan University of Shanghai for Science and Technology School of Optical-Electrical and Computer Engineering Shanghai200093 China

With the increasing number of edited videos, many robust video fingerprinting schemes have been proposed to solve the problem of video content authentication. However, most of them either deal with the temporal and spatial features symmetrically or insufficiently consider the temporal information. In this work, an end-to-end two-branch network toward robust video fingerprinting (RVFNet) is proposed, where the two branches focus on the temporal and spatial information, respectively. The temporal branch aims to comprehensively capture complex motion patterns by combining subtle motion changes with the overall motion trend. The spatial branch exploits the pixel-level information obtained by multiple receptive fields while preserving significant structural features. Deep metric learning is employed in the training process, and we adopt hard triplet loss to constrain the generation of fingerprints. Furthermore, we construct a large-scale and complex dataset for the robust video fingerprinting task based on multiple video content-preserving manipulations in actual scenarios. The size of our dataset exceeds most datasets adopted in the current robust video fingerprinting schemes. Based on the proposed dataset, experimental results demonstrate that our scheme achieves outstanding performance improvements compared with the state of the art. © 2020 IEEE.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

Unsupervised anomaly detection and localization via bidirectional knowledge distillation

引用

Neural Computing and Applications 2024年第29期36卷 18499-18514页

作者： Wang, Xiaoming Wang, Yongxiong Pan, Zhiqun Wang, Guangpeng School of Optical-Electrical and Computer Engineering University of Shanghai for Science and Technology Shanghai200093 China

Knowledge distillation has demonstrated significant potential in addressing the challenge of unsupervised anomaly detection (AD). The representation discrepancy of anomalies in the teacher–student (T-S) model provides evidence for anomaly detection and localization. However, the teacher model is pretrained for classification, while the anomaly scores in the distillation-based anomaly detection method are indirectly derived from the classification scores. The mismatch between the two tasks can hinder the optimization of the model. To tackle this issue, we propose an innovative bidirectional knowledge distillation model. In this approach, forward knowledge distillation is pivotal in bolstering the model’s capacity for generalization. Simultaneously, backward knowledge distillation promotes diversity in representing anomalies. This reciprocal knowledge exchange effectively wards off potential performance declines due to target inconsistency. Through bidirectional knowledge distillation, we establish a more encompassing and resilient framework for knowledge transfer. Additionally, we introduce a novel data augmentation strategy to simulate anomalies and effectively eliminate unnecessary noise. In experiments on the MVTec AD, the proposed model achieves competitive results compared to state-of-the-art methods, 97.47% on image-level AUC, 98.23% on pixel-level AUC, and 94.77% on instance-level PRO. These results demonstrate the practicality of our approach in anomaly detection and localization. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.

关键词： Anomaly detection

来源：评论

学校读者我要写书评

暂无评论

Detecting Inaudible Voice Commands via Acoustic Attenuation by Multi-channel Microphones

引用

IEEE Transactions on Dependable and Secure Computing 2024年 1-17页

作者： Ji, Xiaoyu Zhang, Guoming Li, Xinfeng Qu, Gang Cheng, Xiuzhen Xu, Wenyuan Electrical Engineering of Zhejiang University Hangzhou China School of Computer Science and Technology of Shandong University Qingdao China Department of Electrical and Computer Engineering at the University of Maryland College Park MD USA

DolphinAttacks (i.e., inaudible voice commands) modulate audible voices over ultrasounds to inject malicious commands silently into voice assistants and manipulate controlled systems (e.g., doors or smart speakers). Eliminating DolphinAttacks is challenging if ever possible since it requires to modify the microphone hardware. In this paper, we design EarArray, a lightweight method that can not only detect such attacks but also identify the direction of attackers without requiring any extra hardware or hardware modification. Essentially, inaudible voice commands are modulated on ultrasounds that inherently attenuate faster than the one of audible sounds. By inspecting the command sound signals via the built-in multiple microphones on smart devices, EarArray is able to estimate the attenuation rate and thus detect the attacks. We propose a model of the propagation of audible sounds and ultrasounds from the sound source to a voice assistant, e.g., a smart speaker, and illustrate the underlying principle and its feasibility. We implemented EarArray using two specially-designed microphone arrays and our experiments show that EarArray can detect inaudible voice commands with an accuracy of above 99% and recognize the direction of the attackers with an accuracy of 97.89% and can also detect the laser-based attack with an accuracy of 100%. IEEE

关键词： Microphones

来源：评论

学校读者我要写书评

暂无评论

Circular super-Gaussian Airy beams

引用

Optics Letters 2025年第10期50卷 3401-3404页

作者： Geng, Tao Wang, Lihui School of Optical-Electrical and Computer Engineering University of Shanghai for Science and Technology Shanghai200093 China

By replacing the exponential decay function in the circular Airy beam (CAB) with a super-Gaussian function, we propose a novel abruptly autofocusing beam, the circular super-Gaussian Airy beam (CSGAB). Similar to CAB, CSGAB has a simple and clear mathematical expression. By selecting appropriate parameters, the peak focal intensity of CSGAB can reach 5.3 times that of CAB. The ratio of the maximum intensity at the focal plane to that at the initial plane reaches 1825, which is 38.7 times that of CAB. © 2025 Optica Publishing Group. All rights, including for text and data mining (TDM), Artificial Intelligence (AI) training, and similar technologies, are reserved.

关键词： Focusing

来源：评论

学校读者我要写书评

暂无评论

Halo reduction multi-exposure image fusion technique

引用

Multimedia Tools and Applications 2025年第13期84卷 12347-12370页

作者： Sharif, Rizwan Amin, Benish Sukhia, Komal Nain Department of Electrical Engineering Institute of Space Technology Islamabad Pakistan Department of Computer Science Institute of Space Technology Islamabad Pakistan

Multi-exposure image fusion (MEF) involves combining images captured at different exposure levels to create a single, well-exposed fused image. MEF has a wide range of applications, including low light, low contrast, night photography, medical imaging, and remote sensing. However, MEF methods often face issues like artifacts, halos around edges, color inconsistencies, noise amplification, and difficulty in preserving fine details. Moreover, assessing the quality of fused images objectively is complex due to the subjective nature of human perception. Solving the challenges is essential to developing efficient MEF techniques that produce high-quality results across various scenarios. The proposed technique introduces an approach to handling halo artifacts and implementing MEF. The Dense Scale-Invariant Feature Transform (DSIFT) is used to capture vital information about image brightness, texture, and edges from source images. Three weight maps are computed from the local mean, signal strength, and the global gradient for initial weight estimation. The local mean represents the brightness of specific image areas, signal strength preserves essential details like textures and edges while reducing image noise, and global gradient helps identify regions with significant pixel value shifts across multiple exposure images. The weight maps are then combined using a weighted average and a Gaussian smoothing filter is applied to reduce inherent noise and discontinuities in the original weights. Subsequently, pyramid decomposition is performed to generate a fused image. The efficiency of the proposed method is extensively tested on challenging multi-exposure image sequences. The results of the proposed approach demonstrate its superiority in both subjective evaluation and objective metrics like the MEF-Structural Similarity Index (MEF-SSIM), Natural Image Quality Evaluator (NIQE) and Gradient based performance measure (QAB/f). © The Author(s), under exclusive licence to Springer

关键词： Image fusion

来源：评论

学校读者我要写书评

暂无评论

Pseudoinvertible Neural Networks

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2024年第2期5卷 602-612页

作者： Bolluyt, Elijah D. Comaniciu, Cristina Stevens Institute of Technology Department of Electrical and Computer Engineering HobokenNJ07030 United States

This work introduces novel architecture components and training procedures to create augmented neural networks with the ability to process data bidirectionally via an end-to-end approximate inverse. We develop pseudoinvertible neural network (PsI-NN) layers which function as drop-in replacements for corresponding convolutional and fully connected layers;by using these, existing architectures gain a pseudoinverse function which, with training, approximately reverses the forward function. For cases where learning both a task and its inverse are necessary or desirable, we show that PsI-NN enabled models match or exceed the quality of results generated by systems that use two separate models while drastically reducing system parameter count. We demonstrate this on two tasks: unpaired image translation and semisupervised classification. In both, PsI-NN greatly reduces parameter count without any loss of output quality;in semisupervised image classification, PsI-NN improves classification performance beyond the baseline autoencoder method. Our approach creates pseudoinvertible versions of existing architectures, circumventing the stringent constraints required to create a true inverse while allowing a single neural network to learn a task in two directions simultaneously. © 2023 IEEE.

关键词： Supervised learning

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：