检索结果-内蒙古大学图书馆

Software engineering and Computer Science (CSECS), International Conference on

作者： Chikwendu Ijeoma Amuche Xiaoling Zhang Chiagoziem Chima Ukwuoma Chikwendu Okechuwku Chinedum Ejiyi Chukwuebuka Joseph School of Information and Communication Engineering dept. University of Electronic Science and Technology of China Chengdu China Sichuan Engineering Technology Research Center for Industrial Internet Intelligent Monitoring and Application Chengdu University of Technology Sichuan P.R. China Department of Biochemistry Federal University of Technology Owerri. Ihiagwa Nigeria College of Nuclear Technology and Automation Engineering Chengdu University of Technology Sichuan P.R. China

ISBN: (数字)9798331522216

ISBN: (纸本)9798331522223

Detecting fraud in multi-relational graphs is a significant difficulty due to the complex nature of fraudulent activities and the inadequacies of conventional Graph Neural Networks (GNNs) in managing heterophilic relationships and long-range dependencies. Current GNN-based models experience over-smoothing and over-squashing, which constrains their capacity to differentiate between fraudulent and legitimate interactions. To tackle these difficulties, we offer MR-GT, a Graph Transformer-based model that integrates multi-relational attention and edge-aware techniques to enhance fraud detection. MR-GT implements edge-aware message passing and edge-based attention bias, enabling the model to concentrate on significant transactional links while disregarding less pertinent interactions. The approach utilizes relation-aware aggregation to accurately identify fraud-related patterns in diverse interactions. Experimental findings on the Yelp and Amazon datasets indicate that MR-GT significantly surpasses state-of-the-art approaches, attaining an AVC of 92.33 on Yelp and 98.25 on Amazon. The ablation study further illustrates the significance of each component, indicating that MR-GT provides a scalable and efficient approach for identifying fraud in extensive, multi-relational graphs.

关键词： Measurement Computer science Attention mechanisms Image edge detection Message passing Focusing Transformers Graph neural networks Fraud Software engineering

来源：评论

学校读者我要写书评

暂无评论

Disorders in fluid filled pipeline structure with elastic helmholtz resonators 29

Disorders in fluid filled pipeline structure with elastic he...

引用

2022 29th International Conference on nuclear engineering, ICONE 2022

作者： Zeng, Qingna Wang, Donghui Zang, Fenggang Zhang, Yixiong Wang, Bihao Yuan, Zhihao Science and Technology on Reactor System Design Technology Laboratory Chengdu610213 China Nuclear Power Institute of China Chengdu610213 China College of Shipbuilding Engineering Harbin Engineering University Harbin150001 China

ISBN: (纸本)9780791886359

This paper studies the influence of hybrid disorders on acoustic transmission characteristic in pipeline with periodic elastic HRs. Band gap structures (BGs) for infinite period are investigated by transfer matrix method and finite element method respectively, and fit well with the corresponding transmission loss (TL) for finite regular periods. By introducing a certain degree of uncertainty level, Interval analysis method is used to convert uncertain parameter problems into two deterministic models. Frequency response function (FRF) for near-periodic structure with single disorder and BGs for quasiperiodic structure with multiple disorders considering cavity radius and periodic length are investigated. The existence of disorders in periodic structure would always reduce attenuation interval and intensity, some disturbance would even change the original acoustic transmission characteristic and excite completely new attenuation zones. The sensitivity of pipeline acoustic vibration transfer characteristic to disorder parameter depends on attenuation mechanism, as for periodic elastic HRs pipeline, acoustic propagation stability is more sensitive to characteristic of locally resonant unit. The work in this paper enriches the stability analysis of pipeline with elastic HRs, which would provide theoretical reference for noise and vibration reduction design in practical engineering applications. © 2022 American Society of Mechanical Engineers (ASME). All rights reserved.

关键词： Transfer matrix method

来源：评论

学校读者我要写书评

暂无评论

Intelligent Strawberry Ripeness and Disease Detection Classification System Based on Deep Learning

Intelligent Strawberry Ripeness and Disease Detection Classi...

引用

International Conference on Electronics and Information technology (EIT)

作者： Jiajun Li Xin Yang Songcheng Qian Junjie Xu Lei Sun Zhaoning Zeng School of Automation Engineering University of Electronic Science and Technology of China Chengdu China College of Energy Chengdu University of Technology Chengdu China School of Earth Science Yunnan University China

ISBN: (数字)9798350369151

ISBN: (纸本)9798350369168

In modern agricultural production, timely detection of crop diseases and accurate assessment of ripeness are crucial for reducing yield loss and enhancing product quality. Strawberries, being a widely popular and economically valuable fruit, are particularly sensitive to diseases and strict in ripeness requirements, making quality control during their production critical. This paper develops a classification system for strawberry ripeness and pest disease detection based on the YOLOv5 deep learning model and automated robotic arm technology. The YOLOv5 model, known for its high efficiency, accurately identifies strawberries and assesses their ripeness and pest conditions, thereby enhancing the precision and speed of detection. The automated robotic arm technology utilizes the detection results from the YOLOv5 model to perform efficient and accurate automated sorting and grading of strawberries, significantly reducing labor costs and human error. By integrating deep learning and automation technologies, this system substantially increases strawberry yield and quality, alleviates labor burdens, and enhances agricultural production efficiency.

关键词： YOLO Deep learning automation Costs Accuracy Production Manipulators Robots Diseases Sorting

来源：评论

学校读者我要写书评

暂无评论

Improved Blending Attention Mechanism in Visual Question Answering

引用

Computer Systems Science & engineering 2023年第10期47卷 1149-1161页

作者： Siyu Lu Yueming Ding Zhengtong Yin Mingzhe Liu Xuan Liu Wenfeng Zheng Lirong Yin School of Automation University of Electronic Science and Technology of ChinaChengdu610054China College of Resource and Environment Engineering Guizhou UniversityGuiyang550025China School of Data Science and Artificial Intelligence Wenzhou University of TechnologyWenzhou325000China School of Public Affairs and Administration University of Electronic Science and Technology of ChinaChengdu611731China Department of Geography and Anthropology Louisiana State UniversityBaton Rouge70803LAUSA

Visual question answering(VQA)has attracted more and more attention in computer vision and natural language *** are committed to studying how to better integrate image features and text features to achieve better results in VQA *** of all features may cause information redundancy and heavy computational *** mechanism is a wise way to solve this ***,using single attention mechanism may cause incomplete concern of *** paper improves the attention mechanism method and proposes a hybrid attention mechanism that combines the spatial attention mechanism method and the channel attention mechanism *** the case that the attention mechanism will cause the loss of the original features,a small portion of image features were added as *** the attention mechanism of text features,a selfattention mechanism was introduced,and the internal structural features of sentences were strengthened to improve the overall *** results show that attention mechanism and feature compensation add 6.1%accuracy to multimodal low-rank bilinear pooling network.

关键词： Visual question answering spatial attention mechanism channel attention mechanism image feature processing text feature extraction

来源：评论

学校读者我要写书评

暂无评论

A Compton Scattering Background Subtraction Method of Gamma Energy Spectrum Based on Gaussian Function Convolution

SSRN

引用

SSRN 2024年

作者： Qin, Ru-Lan Li, Chang-Yuan Qin, Zhang-Jian Zhang, Zhi-Hong Cai, Jun Shanghai Institute of Applied Physics Chinese Academy of Sciences Shanghai201800 China University of Chinese Academy of Sciences Beijing100049 China College of Nuclear Technology and Automation Engineering Chengdu University of Technology Chengdu610059 China

Due to the influence of the Compton effect and the relatively close gamma energy emitted during certain nuclide decays, gamma-ray spectrum analysis often faces challenges, such as a high count of Compton platforms and overlapping peaks, leading to some errors in energy spectrum analysis and low nuclide identification rates. This study proposes a spectrum analysis technique that uses Gaussian convolution to effectively eliminate background counts induced by Compton scattering. To minimize calculation errors in variance, the variance error is decreased from 2.98% to 0.84% through the utilization of a binary method for adjusting the transformation scale. Simulation results show that the average relative error of peak area calculations is less than 10.78%, and the differences between the standard peak center position and the peak center position measured by the Gaussian convolution method are less than 0.049% across a range of Signal-to-Noise Ratio(SNR) levels. These findings demonstrate the Gaussian convolution method's significant noise reduction capabilities and computational stabilities. In the end, empirical spectrum experiments were carried out to confirm that the average relative error of peak area calculation is around 7.63%, and the relative error between the Gaussian convolution method and Gamma Vision peak-seeking result is around 0.18%. The results demonstrate that this method can effectively eliminate the Compton scattering background counts in the gamma spectrum, obtain the peak position and peak area of the full energy peak, and facilitate the accurate identification of the weak peak in the gamma spectrum. © 2024, The Authors. All rights reserved.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

Study on gamma spectrum wavelet noise reduction based on scintillation detector

Study on gamma spectrum wavelet noise reduction based on sci...

引用

2021 3rd International Conference on Applied Machine Learning and Data Science, ICAMLDS 2021

作者： Yu, Yingchi Ge, Liangquan Li, Fei Hu, Chuanhao College of Nuclear Technology and Automation Engineering Chengdu University of Technology China Applied Nuclear Technology in Geosciences Key Laboratory of Sichuan Province China

Wavelet transform is a time-scale signal analysis method, which holds the characteristics of the same window size, arbitrary shape change and multi-resolution analysis. Most applications of wavelet transform need to be based on signal decomposition and reconstruction, that is, through the combination of low-pass and high-pass filters to achieve frequency division, the signal is decomposed into signal detail components and large-scale approximate components. In the measurement process of the original energy spectrum data, the denoting process plays a role in reducing the incidence of statistical fluctuations to a certain extent, which is of profound significance to improve the data accuracy. In this paper, wavelet transform is used as the theoretical basis to denoise the actual energy spectrum, at the same time based on the MATLAB platform, simulation of signal with noise, using the different parameters on the signal processing. Firstly, the signal-to-noise ratio (SNR) of the hard and soft thresholds to the signal processing under the same conditions was compared, and then the change of the SNR was observed when the number of decomposition layers was changed under the wavelet basis of db3 (soft threshold). The experimental results show that: 1) under the same conditions, the soft threshold method has higher SNR than hard threshold;2) The number of decomposition layers and the elimination moment affect the signal-to-noise ratio and smoothness of the processed signal respectively, and when the number of decomposition layers is 7-8 layers, the signal-to-noise ratio is about 29, the signal processing effect is the best. This study has shown that Wavelet method is an effective way to process energy spectrum. © Published under licence by IOP Publishing Ltd.

关键词： Signal to noise ratio

来源：评论

学校读者我要写书评

暂无评论

Implementation of digital constant fraction timing based on field programmable gate array

Implementation of digital constant fraction timing based on ...

引用

2021 3rd International Conference on Applied Machine Learning and Data Science, ICAMLDS 2021

作者： Ding, Wenwan Hu, Chuanhao Cao, Yang Li, Xuegang Yue, Tao Zhao, Kunling Zhang, Song Zhang, Wenbing Gu, Yi College of Nuclear Technology and Automation Engineering Chengdu University of Technology China Applied Nuclear Technology in Geosciences Key Laboratory of Sichuan Province China

Compared with analog circuits, FPGA(Field Programmable Gate Array) had the advantages of redevelopment and low cost. Compared with MCU, FPGA used hardware description language to construct the circuit, which had the advantages of simple structure and fast parallel operation speed. In this paper, the simulation of digital Constant Fraction Timing can be realized based on FPGA, and the feasibility and accuracy of the design can be proved by using the simulation waveform of Modelsim and the waveform can be captured by signaltap. Modelsim can verify the correctness of the timing between modules, and signaltap can verify the actual running effect of the program on the board. © Published under licence by IOP Publishing Ltd.

关键词： Computer hardware description languages

来源：评论

学校读者我要写书评

暂无评论

Spectral data processing based on maximum noise fraction

Spectral data processing based on maximum noise fraction

引用

2021 3rd International Conference on Applied Machine Learning and Data Science, ICAMLDS 2021

作者： Guo, Wentai Ge, Liangquan Li, Fei Hu, Chuanhao College of Nuclear Technology and Automation Engineering Chengdu University of Technology China Applied Nuclear Technology in Geosciences Key Laboratory of Sichuan Province China

The processing of gamma spectrum measurement data has always been an important step in data interpretation of the survey area. It includes the correction of radioactivity interference caused by non-measuring targets, the correction of effects caused by different heights or topography in the survey environment, the smoothness and denoising of the measurement spectrum, etc. The accuracy of each step will affect the final data resources. The interpretation of the materials results in inaccurate regional radioactivity assessment. Maximum noise separation transformation is a similar method to principal component transformation, which is used to reduce dimensionality and noise of data. This paper takes the noise reduction process of gamma spectrum data processing as the research direction, uses SPSS data processing software, combines with MATLAB to denoise the data, and uses the maximum noise separation transformation as the denoising method, applies it to denoise the spectral data, and uses the principal component analysis method and the maximum noise fraction method to denoise the same set of data separately. After analysis, it is concluded that the maximum noise fraction method is superior to principal component analysis when there is a large noise variance for a set of data. © Published under licence by IOP Publishing Ltd.

关键词： Spectrum analysis

来源：评论

学校读者我要写书评

暂无评论

Towards in X-Ray Induced Acoustic Imaging for Nasopharyngeal Carcinoma Radiotherapy with a Multidimensional Information Fusion Network

Towards in X-Ray Induced Acoustic Imaging for Nasopharyngeal...

引用

International Conference on Computer Applications technology (ICCAT)

作者： Xin Jiang Zhi-Chao Chen Zheng Sun Ming-Zhe Liu School of Artificial Intelligence Chongqing University of Technology Chongqing China Department of Radiology Chongqing University Cancer Hospital Chongqing China College of Nuclear Technology and Automation Engineering Chengdu University of Technology Chengdu China School of Data Science and Artificial Intelligence Wenzhou University of Technology Wenzhou China

ISBN: (数字)9798331539795

ISBN: (纸本)9798331539801

X-ray-induced acoustic computed tomography (XACT) imaging is an emerging noninvasive imaging technique with wide applications in various clinical medicine fields. However, traditional XACT imaging methods often suffer from low imaging quality and increased noise artifacts, particularly notable under sparse sampling conditions. To overcome these challenges, we propose a hybrid supervised reconstruction method based on deep learning. Specifically, we extract and integrate information from XACT thermoacoustic signals and XACT images reconstructed by traditional methods to leverage complementary feature information from both modalities. To adapt to imaging under sparse sampling, we define multiple sparse conditions based on real-world imaging environments and validate them using a real radiotherapy dataset of head and neck cancer patients. The results demonstrate the adaptability of our method to sparse conditions and its capability to enhance XACT imaging quality under such circumstances. Additionally, we validate the feasibility of X-ray thermoacoustic imaging for 3D dose monitoring.

关键词： Deep learning Three-dimensional displays Imaging Feature extraction Acoustics Robustness Radiation therapy Data mining X-ray imaging Image reconstruction

来源：评论

学校读者我要写书评

暂无评论

Analysis and Design of Finite Input Inductance Class-E Inverters with Variable Complex Loads

Analysis and Design of Finite Input Inductance Class-E Inver...

引用

IEEE Workshop on Computers in Power Electronics

作者： Yufei Wang Min Sun Dongsheng Cai Jian Li Qi Huang Siming Pan School of Mechanical and Electrical Engineering University of Electronic Science and Technology of China Chengdu China Institute of Electronic and Information Engineering of UESTC in Guangdong Dongguan China College of Nuclear Technology and Automation Engineering Chengdu University of Technology Chengdu China Southwest University of Science and Technology Mianyang China

This paper presents a detailed analysis and design method of the Class-E inverter with finite input inductance under arbitrary complex loads based on the first harmonic approximation. The analytic results show that the inverter can achieve ZVS with a wide range of complex loads with proper resonant network design. The model established allows designers to plot the theoretical ZVS load impedance region for specific design parameters directly in the complex plane. Subsequently a design method for the inverter to achieve high efficiency operation under variable complex loads is developed. This method thus allows the Class-E inverter to be used in an application in which the load resistance varies as well as the reactance. The effectiveness of the proposed method is verified by a 13.56 MHz Class-E inverter prototype.

关键词：

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：