检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Wang, Xiao Jin, Yu Wu, Wentao Zhang, Wei Zhu, Lin Jiang, Bo Tian, Yonghong School of Computer Science and Technology Anhui University Hefei China School of Artificial Intelligence Anhui University Hefei China Peng Cheng Laboratory Shenzhen China Beijing Institute of Technology Beijing China National Key Laboratory for Multimedia Information Processing School of Computer Science Peking University China School of Electronic and Computer Engineering Shenzhen Graduate School Peking University China

Object detection in event streams has emerged as a cutting-edge research area, demonstrating superior performance in low-light conditions, scenarios with motion blur, and rapid movements. Current detectors leverage spiking neural networks, Transformers, or convolutional neural networks as their core architectures, each with its own set of limitations including restricted performance, high computational overhead, or limited local receptive fields. This paper introduces a novel MoE (Mixture of Experts) heat conduction-based object detection algorithm that strikingly balances accuracy and computational efficiency. Initially, we employ a stem network for event data embedding, followed by processing through our innovative MoE-HCO blocks. Each block integrates various expert modules to mimic heat conduction within event streams. Subsequently, an IoU-based query selection module is utilized for efficient token extraction, which is then channeled into a detection head for the final object detection process. Furthermore, we are pleased to introduce EvDET200K, a novel benchmark dataset for event-based object detection. Captured with a high-definition Prophesee EVK4-HD event camera, this dataset encompasses 10 distinct categories, 200,000 bounding boxes, and 10,054 samples, each spanning 2 to 5 seconds. We also provide comprehensive results from over 15 state-of-the-art detectors, offering a solid foundation for future research and comparison. The source code of this paper will be released on: https://***/EventAHU/OpenEvDET Copyright © 2024, The Authors. All rights reserved.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Generative Multi-View HRRP Recognition Based on Cascade Generation and Fusion Network

Generative Multi-View HRRP Recognition Based on Cascade Gene...

引用

IEEE International Conference on Radar

作者： Qiang Zhou Bingqian Yu Yanhua Wang Liang Zhang Le Zheng Difan Zou Xin Zhang School of Information and Electronics Beijing Institute of Technology Beijing China Beijing Key Laboratory of Embedded Real-Time Information Processing Technology Beijing China Department of Computer Science University of Hong Kong Hongkong China HKU Musketeers Foundation Institute of Data Science University of Hong Kong Hongkong China

ISBN: (数字)9798350362381

ISBN: (纸本)9798350362398

High-resolution range profile (HRRP) is critical for radar target recognition. However, HRRP data for non-cooperative targets are often sparsely collected, leading to limited performance of HRRP target recognition methods. To address the issue of sparse view target recognition, we propose a generative multi-view HRRP recognition (GMVR) model. Firstly, by leveraging the pattern learning and generation ability of deep generative models (DGMs), we propose a viewpoint feature expansion module (VFEM) to generate multi-view features adjacent to sparse viewpoints. Subsequently, based on feature-level fusion, we proposed a multi-view feature fusion module (MVFM). Aided by attention mechanisms, the generated features can be adaptively integrated, thereby reducing the decision bias caused by single-viewpoint HRRP samples. Additionally, we employ transfer learning to mitigate the small sample problem caused by sparse viewpoints. Experimental results demonstrate that the proposed method outperforms various comparison methods, achieving optimal performance.

关键词： Adaptation models Attention mechanisms Target recognition Transfer learning Radar Pattern recognition

来源：评论

学校读者我要写书评

暂无评论

The classification and retrieval of the image affective semantics based on integration of multi features and svm

引用

Journal of information Hiding and Multimedia Signal processing 2018年第4期9卷 864-873页

作者： Chen, Hui Xu, Lin Zhang, Fu Quan College of Electronics and Information Science Fujian Jiangxia University Room 4#608Jincheng Garden Xindian Fuzhou350012 China Innovative Information Industry Research Institute Fujian Normal University Fuqing Brunch of Fujian Normal University Campus County 1# Longjiang Road Fuqing Fuzhou350300 China School of Computer Science & Technology Beijing Institute of Technology Fujian Provincial Key Laboratory of Information Processing and Intelligent Control Minjiang University Room A9#301 Juyuan Jinshan Jiyuan Fuzhou350008 China

Images contain rich emotions. The traditional image classification and retrieval methods mainly classify images according to the bottom visual features of the images. Most of them ignore the influence and effect of emotions. The image affective classification and retrieval method in this paper combines the bottom visual features with the high-level affective semantics. Aiming at the insufficiency of the retrieval based on the single bottom visual features, it combines the bottom image color feature with the shape feature, and establishes the image feature space. Aiming at the semantic gap between low-level visual features and high-level semantics of the image, it establishes the emotional semantic space through cognitive psychology experiments and factor analysis. The method in this paper uses SVM to fill the semantic blanks to achieve the affective semantics annotation of the image, maps feature space to the emotional space of the image to realize the classification and retrieval of the image affective semantics. The result of the experiment shows that this method improves the efficiency and performance of image retrieval, and provides an effective solution to the affective semantics research of images. © 2018, Ubiquitous International. All rights reserved.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

An Algorithm for Detecting Surface Defects in Industrial Strip Steel based on Receptive Field and Feature information Supplementation

An Algorithm for Detecting Surface Defects in Industrial Str...

引用

International Conference on computer Supported Cooperative Work in Design

作者： Qingyu Xu Jiguo Yu Anming Dong Zihao Shang Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center (National Supercomputer Center in Jinan) Qilu University of Technology (Shandong Academy of Sciences) Jinan China School of Computer Science and Technology Qilu University of Technology (Shandong Academy of Sciences) Jinan China Big Data Institute Qilu University of Technology (Shandong Academy of Sciences) Jinan China Shandong Provincial Key Laboratory of Computer Networks Shandong Fundamental Research Center for Computer Science Jinan China Shandong Data Intelligence Information Technology Co. Ltd. Jinan China

ISBN: (数字)9798350349184

ISBN: (纸本)9798350349191

In the context of Industry 4.0 and the rise of intelligent manufacturing, the quality of industrial products is becoming more and more important. Strip steel surface defect detection, as a key link in industrial production, is crucial to ensure the quality of industrial products. However, due to the irregularity of the defect scale and the inconspicuous defect features in the surface defect image of strip steel, it is difficult for the existing detection algorithms to realize the effective detection of defects. In order to better extract the features of defects and improve the network’s ability to detect defects, this paper proposes an algorithm for detecting surface defects on industrial strip steel based on receptive field and feature information supplementation. First, we design a receptive field (RF) module to replace the residual structure in the C3 module, which we name C3RF. This module can effectively increase the network’s receptive field, allowing the network to fully capture irregular defect features without increasing the cost. Second, for the characteristics that defective features are not obvious and tend to lose detail information as the network deepens, an extra information supplemental branching feature fusion pyramid (EFPN) is proposed on the basis of the original PAFPN architecture to compensate for the detail information that is lost by the fragile features in deeper layers. Finally, convolutional block attention module (CBAM) is introduced to replace the spatial pooling pyramid (SPPF) in the baseline network, which enhances the contrast between defects and backgrounds, and improves the classification and localization ability of the network. Our network achieves an accuracy of 82.2% on the publicly available strip steel defect detection dataset, which is a 4.0% improvement over the baseline. The results show that the network constructed in this paper can realize effective defect detection.

关键词： Radio frequency Location awareness Strips Production Feature extraction Real-time systems Hardware

来源：评论

学校读者我要写书评

暂无评论

A performance evaluation of correspondence grouping methods for 3D rigid data matching

arXiv

引用

arXiv 2019年

作者： Yang, Jiaqi Xian, Ke Wang, Peng Zhang, Yanning School of Computer Science Northwestern Polytechnical University National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology Xi'an710129 China National Key Laboratory of Science and Technology on Multi-spectral Information Processing School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan430074 China

Seeking consistent point-to-point correspondences between 3D rigid data (point clouds, meshes, or depth maps) is a fundamental problem in 3D computer vision. While a number of correspondence selection methods have been proposed in recent years, their advantages and shortcomings remain unclear regarding different applications and perturbations. To fill this gap, this paper gives a comprehensive evaluation of nine state-of-the-art 3D correspondence grouping methods. A good correspondence grouping algorithm is expected to retrieve as many as inliers from initial feature matches, giving a rise in both precision and recall as well as facilitating accurate transformation estimation. Toward this rule, we deploy experiments on three benchmarks with different application contexts including shape retrieval, 3D object recognition, and point cloud registration together with various perturbations such as noise, point density variation, clutter, occlusion, partial overlap, different scales of initial correspondences, and different combinations of keypoint detectors and descriptors. The rich variety of application scenarios and nuisances result in different spatial distributions and inlier ratios of initial feature correspondences, thus enabling a thorough evaluation. Based on the outcomes, we give a summary of the traits, merits, and demerits of evaluated approaches and indicate some potential future research directions. Copyright © 2019, The Authors. All rights reserved.

关键词： Benchmarking

来源：评论

学校读者我要写书评

暂无评论

Multi-view ontology alignment visualization (MOAV): A human-cognition based ontology alignment visualization technique

Journal of Network Intelligence

引用

Journal of Network Intelligence 2020年第4期5卷 198-210页

作者： Chen, Jie Zhang, Jie Xue, Xingsi Huang, Yikun Fujian Provincial Key Laboratory of Big Data Mining and Applications Fujian University of Technology No.33 Xueyuan Road University Town Minhou FuzhouFujian350118 China School of Computer Science and Engineering Yulin Normal University No.299 Education Middle Road Yulin CityGuanxi Province537000 China Fujian Provincial Key Laboratory of Big Data Mining and Applications Intelligent Information Processing Research Center Fujian University of Technology No.33 Xueyuan Road University Town Minhou FuzhouFujian350118 China Guangxi Key Laboratory of Automatic Detecting Technology and Instruments Guilin University of Electronic Technology No.1 Jinji Road Guangxi Guilin541004 China Department of Information Technology Concord University College Fujian Normal University No.68 Xueyuan Road University Town Minhou FuzhouFujian350117 China

To further improve the quality of ontology alignment, it is necessary for an ontology matcher to introduce a user’s knowledge into its automatic matching process, which yields the development of interactive ontology matching techniques. Since validating problematic entity correspondences is a difficult cognition task, user interaction based on Ontology Alignment Visualization (OAV) has become the critical component of an interactive ontology matcher, which directly affects the quality of validating result and the efficiency of validating process. The existing OAV tools are mainly developed based on their designers’ subjective feelings and experiences, which do not take into consideration the law of human cognition. To improve the efficiency of the ontology alignment validating process, in this paper, a Multi-view OAV (MOAV) is proposed, which synthetically utilizes the human cognitive theory-information visualization and human-computer inter-action. The experiment utilizes the Ontology Alignment Evaluation Initiative (OAEI)’s benchmark to test the performance of our proposal, and the experimental results show that MOAV can effectively improve a user’s validating efficiency. © 2020, Taiwan Ubiquitous information. All rights reserved.

关键词： Visualization

来源：评论

学校读者我要写书评

暂无评论

Sigma Set: A small second order statistical region descriptor

Sigma Set: A small second order statistical region descripto...

引用

Conference on computer Vision and Pattern Recognition (CVPR)

作者： Xiaopeng Hong Hong Chang Shiguang Shan Xilin Chen Wen Gao School of Computer Science and Technology Harbin Institute of Technology China Institute of Computing Technology Chinese Academy and Sciences China Key Laboratory of Intelligent Information Processing Chinese Academy and Sciences China Institute of Computing Technology Chinese Academy of Sciences China Institute for Digital Media Peking University China

Given an image region of pixels, second order statistics can be used to construct a descriptor for object representation. One example is the covariance matrix descriptor, which shows high discriminative power and good robustness in many computer vision applications. However, operations for the covariance matrix on Riemannian manifolds are usually computationally demanding. This paper proposes a novel second order statistics based region descriptor, named “Sigma Set”, in the form of a small set of vectors, which can be uniquely constructed through Cholesky decomposition on the covariance matrix. Sigma Set is of low dimension, powerful and robust. Moreover, compared with the covariance matrix, Sigma Set is not only more efficient in distance evaluation and average calculation, but also easier to be enriched with first order statistics. Experimental results in texture classification and object tracking verify the effectiveness and efficiency of this novel object descriptor.

关键词： Robustness Histograms Fractals Lighting Statistics Solids computer science Power engineering and energy Mathematics Power engineering computing

来源：评论

学校读者我要写书评

暂无评论

Deep learning based joint resource scheduling algorithms for hybrid MEC networks

arXiv

引用

arXiv 2019年

作者： Jiang, Feibo Wang, Kezhi Dong, Li Pan, Cunhua Xu, Wei Yang, Kun Hunan Provincial Key Laboratory of Intelligent Computing and Language Information Processing Hunan Normal University Changsha China Department of Computer and Information Sciences Northumbria University United Kingdom Key Laboratory of Hunan Province for New Retail Virtual Reality Technology Hunan University of Technology and Business Changsha China School of Electronic Engineering and Computer Science Queen Mary University of London LondonE1 4NS United Kingdom NCRL Southeast University Nanjing China School of Computer Technology and Engineering Changchun Institute of Technology Changchun China School of Computer Sciences and Electrical Engineering University of Essex ColchesterCO4 3SQ United Kingdom

In this paper, we consider a hybrid mobile edge computing (H-MEC) platform, which includes ground stations (GSs), ground vehicles (GVs) and unmanned aerial vehicle (UAVs), all with mobile edge cloud installed to enable user equipments (UEs) or Internet of thing (IoT) devices with intensive computing tasks to offload. Our objective is to obtain an online offloading algorithm to minimize the energy consumption of all the UEs, by jointly optimizing the positions of GVs and UAVs, user association and resource allocation in real-time, while considering the dynamic environment. To this end, we propose a hybrid deep learning based online offloading (H2O) framework where a large-scale path-loss fuzzy c-means (LSFCM) algorithm is first proposed and used to predict the optimal positions of GVs and UAVs. Secondly, a fuzzy membership matrix U-based particle swarm optimization (U-PSO) algorithm is applied to solve the mixed integer nonlinear programming (MINLP) problems and generate the sample datasets for the deep neural network (DNN) where the fuzzy membership matrix can capture the small-scale fading effects and the information of mutual interference. Thirdly, a DNN with the scheduling layer is introduced to provide user association and computing resource allocation under the practical latency requirement of the tasks and limited available computing resource of H-MEC. In addition, different from traditional DNN predictor, we only input one UEs information to the DNN at one time, which will be suitable for the scenarios where the number of UE is varying and avoid the curse of dimensionality in DNN. Copyright © 2019, The Authors. All rights reserved.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

MEFFGRN: Matrix enhancement and feature fusion-based method for reconstructing the gene regulatory network of epithelioma papulosum cyprini cells by spring viremia of carp virus infection

引用

computers in Biology and Medicine 2024年 179卷 108835-108835页

作者： Wei, Pi-Jing Bao, Jin-Jin Gao, Zhen Tan, Jing-Yun Cao, Rui-Fen Su, Yansen Zheng, Chun-Hou Deng, Li Key Laboratory of Intelligent Computing & Signal Processing of Ministry of Education Institutes of Physical Science and Information Technology Anhui University 111 Jiulong Road HefeiAnhui230601 China Key Laboratory of Intelligent Computing & Signal Processing of Ministry of Education School of Computer Science and Technology Anhui University 111 Jiulong Road HefeiAnhui230601 China Shenzhen Key Laboratory of Microbial Genetic Engineering College of Life Sciences and Oceanology Shenzhen University Guangdong Shenzhen518055 China School of Artificial Intelligence Anhui University 111 Jiulong Road HefeiAnhui230601 China

Gene regulatory networks (GRNs) are crucial for understanding organismal molecular mechanisms and processes. Construction of GRN in the epithelioma papulosum cyprini (EPC) cells of cyprinid fish by spring viremia of carp virus (SVCV) infection helps understand the immune regulatory mechanisms that enhance the survival capabilities of cyprinid fish. Although many computational methods have been used to infer GRNs, specialized approaches for predicting the GRN of EPC cells following SVCV infection are lacking. In addition, most existing methods focus primarily on gene expression features, neglecting the valuable network structural information in known GRNs. In this study, we propose a novel supervised deep neural network, named MEFFGRN (Matrix Enhancement- and Feature Fusion-based method for Gene Regulatory Network inference), to accurately predict the GRN of EPC cells following SVCV infection. MEFFGRN considers both gene expression data and network structure information of known GRN and introduces a matrix enhancement method to address the sparsity issue of known GRN, extracting richer network structure information. To optimize the benefits of CNN (Convolutional Neural Network) in image processing, gene expression and enhanced GRN data were transformed into histogram images for each gene pair respectively. Subsequently, these histograms were separately fed into CNNs for training to obtain the corresponding gene expression and network structural features. Furthermore, a feature fusion mechanism was introduced to comprehensively integrate the gene expression and network structural features. This integration considers the specificity of each feature and their interactive information, resulting in a more comprehensive and precise feature representation during the fusion process. Experimental results from both real-world and benchmark datasets demonstrate that MEFFGRN achieves competitive performance compared with state-of-the-art computational methods. Furthermore, study

关键词： Gene expression

来源：评论

学校读者我要写书评

暂无评论

Combined Representation Learning for Uncertain Knowledge Graphs

Combined Representation Learning for Uncertain Knowledge Gra...

引用

Mechatronics, IoT and Industrial Informatics (ICMIII), International Conference on

作者： Rui Qiao Xiaolei Zhong Yang Fu School of Computer Science and Technology Wuhan University of Science and Technology Wuhan China Hubei Province Key Laboratory of Intelligent Information Processing and Real-Time Industrial System Wuhan University of Science and Technology Wuhan China Dianchi College Kunming China Huazhong University of Science and Technology Wuhan China

ISBN: (数字)9798350386639

ISBN: (纸本)9798350386646

Uncertain knowledge graphs store the relations between entities in the real world in the form of weighted triples, where the weights indicate the confidence scores of the triples to be true. However, there is limited research on the inference of uncertain knowledge graphs, especially in the integration of symbolic and numerical methods. Rules are explicit knowledge. Encoding the rules directly can provide more knowledge graph semantic information and can make full use of the interpretability and accuracy of logical rules. Unfortunately, none of the existing methods directly encode rules for uncertain knowledge graphs, but only use rules to calculate the confidence of unseen triples in training data. In this paper, we propose a new reasoning model CRLUKG for uncertain knowledge graphs, that combines the advantages of symbolic and numerical reasoning. We directly encode rules mined from the knowledge graph as one of the inputs of the CRLUKG model. Compositional representation learning with triple and rule embedding is implemented for optimizing objective function for triples and relations pairs. Experimental results demonstrate that our proposed method enhances the accuracy of reasoning, and furthermore, CRLUKG exhibits superior performance in comparison to baseline models when handling multi-task scenarios.

关键词： Representation learning Accuracy Large language models Semantics Training data Knowledge graphs Tail

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：