检索结果-南通市图书馆

A Deep Model for Partial Multi-label Image Classification with Curriculum-based Disambiguation

在线全文

学校读者我要写书评

暂无评论

Machine Intelligence Research 2024年第4期21卷 801-814页

作者： feng Sun Ming-Kun Xie Sheng-Jun Huang MIIT Key Laboratory of Pattern Analysis and Machine Intelligence College of Computer Science and TechnologyNanjing University of AeronauticsandAstronauticsNanjing211106China

In this paper,we study the partial multi-label(PML)image classification problem,where each image is annotated with a candidate label set consisting of multiple relevant labels and other noisy *** PML methods typically design a disambiguation strategy to filter out noisy labels by utilizing prior knowledge with extra assumptions,which unfortunately is unavailable in many real ***,because the objective function for disambiguation is usually elaborately designed on the whole training set,it can hardly be optimized in a deep model with stochastic gradient descent(SGD)on *** this paper,for the first time,we propose a deep model for PML to enhance the representation and discrimination *** the one hand,we propose a novel curriculum-based disambiguation strategy to progressively identify ground-truth labels by incorporating the varied difficulties of different *** the other hand,consistency regularization is introduced for model training to balance fitting identified easy labels and exploiting potential relevant *** experimental results on the commonly used benchmark datasets show that the proposed method significantlyoutperforms the SOTA methods.

关键词： Partial multi-label image classification curriculum-based disambiguation consistency regularization label difficulty candidatelabel set.

Improving the Short-Range Precipitation Forecast of Numerical Weather Prediction through a Deep Learning-Based Mask Approach

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Advances in Atmospheric Sciences 2024年第8期41卷 1601-1613页

作者： Jiaqi ZHENG Qing LING Jia LI Yerong feng School of Computer Science and Engineering/Guangdong Provincial Key Laboratory of Computational Science Sun Yat-Sen UniversityGuangzhou 510006China School of Mathematics Sun Yat-Sen UniversityGuangzhou 510275China Guangzhou Institute of Tropical and Marine Meteorology/Guangdong Provincial Key Laboratory of Regional Numerical Weather Prediction Guangzhou 510641China Shenzhen Institute of Meteorological Innovation Shenzhen 518038China

Due to various technical issues,existing numerical weather prediction(NWP)models often perform poorly at forecasting rainfall in the first several *** correct the bias of an NWP model and improve the accuracy of short-range precipitation forecasting,we propose a deep learning-based approach called UNet Mask,which combines NWP forecasts with the output of a convolutional neural network called *** UNet Mask involves training the UNet on historical data from the NWP model and gridded rainfall observations for 6-hour precipitation *** overlap of the UNet output and the NWP forecasts at the same rainfall threshold yields a *** UNet Mask blends the UNet output and the NWP forecasts by taking the maximum between them and passing through the mask,which provides the corrected 6-hour rainfall *** evaluated UNet Mask on a test set and in real-time *** results showed that UNet Mask outperforms the NWP model in 6-hour precipitation prediction by reducing the FAR and improving CSI *** tests also showed that different small rainfall thresholds applied to the UNet and the NWP model have different effects on UNet Mask's forecast *** study shows that UNet Mask is a promising approach for improving rainfall forecasting of NWP models.

关键词： deep learning numerical weather prediction(NWP) 6-hour quantitative precipitation forecast

A versatile model with three-dimensional triangular lattice for unconventional transport and various topological effects

在线全文

学校读者我要写书评

暂无评论

National Science Review 2024年第1期11卷 204-214页

作者： Jing-Yang You Gang Su Yuan Ping feng Department of Physics National University of Singapore Kavli Institute for Theoretical Sciences and CAS Center for Excellence in Topological Quantum Computation University of Chinese Academy of Sciences Centre for Advanced 2D Materials National University of Singapore

The finite Berry curvature in topological materials can induce many subtle phenomena, such as the anomalous Hall effect(AHE), spin Hall effect(SHE), anomalous Nernst effect(ANE), non-linear Hall effect(NLHE) and bulk photovoltaic effects. To explore these novel physics as well as their connection and coupling, a precise and effective model should be developed. Here, we propose such a versatile model—a 3D triangular lattice with alternating hopping parameters, which can yield various topological phases,including kagome bands, triply degenerate fermions, double Weyl semimetals and so on. We reveal that this special lattice can present unconventional transport due to its unique topological surface states and the aforementioned topological phenomena, such as AHE, ANE, NLHE and the topological photocurrent effect. In addition, we also provide a number of material candidates that have been synthesized experimentally with this lattice, and discuss two materials, including a non-magnetic triangular system for SHE, NLHE and the shift current, and a ferromagnetic triangular lattice for AHE and ANE. Our work provides an excellent platform, including both the model and materials, for the study of Berry-curvature-related physics.

关键词： three-dimensional triangular lattice different topological phases various topological effects Berry curvature

Design and implementation of a multi-tile parallel scanning rasterization accelerator

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

The Journal of China Universities of Posts and Telecommunications 2024年第2期31卷 94-104页

作者： Xing Lidong Guo Qiang Peng Xinlong feng Zhenfu School of Electronic Engineering Xi'an University of Posts and TelecommunicationsXi'an 710121China

In the design of a graphic processing unit(GPU),the processing speed of triangle rasterization is an important factor that determines the performance of the *** architecture of a multi-tile parallel-scan rasterization accelerator was proposed in this *** accelerator uses a bounding box algorithm to improve scanning *** rasterizes multiple tiles in parallel and scans multiple lines at the same time within each *** highly parallel approach drastically improves the performance of *** the 65 nm process standard cell library of Semiconductor Manufacturing International Corporation(SMIC),the accelerator can be synthesized to a maximum clock frequency of 220 *** implementation on the Genesys2 field programmable gate array(FPGA)board fully verifies the functionality of the *** implementation shows a significant improvement in rendering speed and efficiency and proves its suitability for high-performance rasterization.

关键词： graphic processing unit(GPU) rasterization multi-tile parallelism

A Novel Multi-Stream Fusion Network for Underwater Image Enhancement

在线全文

学校读者我要写书评

暂无评论

China Communications 2024年第2期21卷 166-182页

作者： Guijin Tang Lian Duan Haitao Zhao feng Liu Jiangsu Key Laboratory of Image Processing and Image Communication Nanjing University of Posts and TelecommunicationsNanjing 210003China Jiangsu Key Laboratory of Wireless Communications Nanjing University of Posts and TelecommunicationsNanjing 210003China

Due to the selective absorption of light and the existence of a large number of floating media in sea water, underwater images often suffer from color casts and detail blurs. It is therefore necessary to perform color correction and detail restoration. However,the existing enhancement algorithms cannot achieve the desired results. In order to solve the above problems, this paper proposes a multi-stream feature fusion network. First, an underwater image is preprocessed to obtain potential information from the illumination stream, color stream and structure stream by histogram equalization with contrast limitation, gamma correction and white balance, respectively. Next, these three streams and the original raw stream are sent to the residual blocks to extract the features. The features will be subsequently fused. It can enhance feature representation in underwater images. In the meantime, a composite loss function including three terms is used to ensure the quality of the enhanced image from the three aspects of color balance, structure preservation and image smoothness. Therefore, the enhanced image is more in line with human visual ***, the effectiveness of the proposed method is verified by comparison experiments with many stateof-the-art underwater image enhancement algorithms. Experimental results show that the proposed method provides superior results over them in terms of MSE,PSNR, SSIM, UIQM and UCIQE, and the enhanced images are more similar to their ground truth images.

关键词： image enhancement multi-stream fusion underwater image

Machine learning in metal-ion battery research: Advancing material prediction, characterization, and status evaluation

在线全文

学校读者我要写书评

暂无评论

Journal of Energy Chemistry 2024年第3期90卷 191-204,I0006页

作者： Tong Yu Chunyang Wang Huicong Yang feng Li Shenyang National Laboratory for Materials Science Institute of Metal ResearchChinese Academy of SciencesShenyang 110016LiaoningChina School of Materials Science and Engineering University of Science and Technology of ChinaShenyang 110016LiaoningChina Department of Physics and Astronomy University of California IrvineIrvineCA92697USA

Metal-ion batteries(MIBs),including alkali metal-ion(Li^(+),Na^(+),and K^(3)),multi-valent metal-ion(Zn^(2+),Mg^(2+),and Al^(3+)),metal-air,and metal-sulfur batteries,play an indispensable role in electrochemical energy ***,the performance of MIBs is significantly influenced by numerous variables,resulting in multi-dimensional and long-term challenges in the field of battery research and performance *** learning(ML),with its capability to solve intricate tasks and perform robust data processing,is now catalyzing a revolutionary transformation in the development of MIB materials and *** this review,we summarize the utilization of ML algorithms that have expedited research on MIBs over the past five *** present an extensive overview of existing algorithms,elucidating their details,advantages,and limitations in various applications,which encompass electrode screening,material property prediction,electrolyte formulation design,electrode material characterization,manufacturing parameter optimization,and real-time battery status ***,we propose potential solutions and future directions for the application of ML in advancing MIB development.

关键词： Metal-ion battery Machine learning Electrode materials Characterization Status evaluation

Maximum Principle for Stochastic Control System with Elephant Memory and Jump Diffusion

在线全文

学校读者我要写书评

暂无评论

Journal of Systems Science & Complexity 2024年第4期37卷 1392-1412页

作者： feng Siqi GAO Lei WANG Guangchen XIAO Hua School of Control Science and Engineering Shandong UniversityJinan 250061China School of Economics Shandong UniversityJinan 250100China School of Mathematics and Statistics Shandong UniversityWeihai 264209China

Motivated by a duopoly game problem,the authors study an optimal control problem where the system is driven by Brownian motion and Poisson point process and has elephant memory for the control variable and the state ***,the authors establish the unique solvability of an anticipated backward stochastic differential equation,derive a stochastic maximum principle,and prove a verification theorem for the aforementioned optimal control ***,the authors generalize these results to nonzero-sum stochastic differential game ***,the authors apply the theoretical results to the duopoly game problem and obtain the corresponding Nash equilibrium solution.

关键词： Anticipated backward stochastic differential equation elephant memory Fréchet derivative maximum principle stochastic delayed differential equation

Health diagnosis of ultrahigh arch dam performance using heterogeneous spatial panel vector model

在线全文

学校读者我要写书评

暂无评论

Water Science and Engineering 2024年第2期17卷 177-186页

作者： Er-feng Zhao Xin Li Chong-shi Gu The National Key Laboratory of Water Disaster Prevention Hohai UniversityNanjing 210098China NationalEngineering Research Centerof Water Resources Eficient Utilization and Engineering Safety Hohai UniversityNanjing 20098China

Currently,more than ten ultrahigh arch dams have been constructed or are being constructed in *** control is essential to long-term operation of these *** study employed the flexibility coefficient and plastic complementary energy norm to assess the structural safety of arch dams.A comprehensive analysis was conducted,focusing on differences among conventional methods in characterizing the structural behavior of the Xiaowan arch dam in ***,the spatiotemporal characteristics of the measured performance of the Xiaowan dam were explored,including periodicity,convergence,and time-effect *** findings revealed the governing mechanism of main ***,a heterogeneous spatial panel vector model was developed,considering both common factors and specific factors affecting the safety and performance of arch *** model aims to comprehensively illustrate spatial heterogeneity between the entire structure and local regions,introducing a specific effect quantity to characterize local deformation ***,the proposed model was applied to the Xiaowan arch dam,accurately quantifying the spatiotemporal heterogeneity of dam ***,the spatiotemporal distri-bution characteristics of environmental load effects on different parts of the dam were reasonably *** of the model prediction enhances its credibility,leading to the formulation of health diagnosis criteria for future long-term operation of the Xiaowan *** findings not only enhance the predictive ability and timely control of ultrahigh arch dams'performance but also provide a crucial basis for assessing the effectiveness of engineering treatment measures.

关键词： Ultrahigh arch dam Structural performance Deformation behavior Diagnosis criterion Panel data model

Alignment of Track Oscillations during Tropical Cyclone Rapid Intensification

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Advances in Atmospheric Sciences 2024年第4期41卷 655-670页

作者： Tong XIE Liguang WU Yecheng feng Jinghua YU Key Laboratory of Meteorological Disaster of Ministry of Education Nanjing University of Information Science and TechnologyNanjing 210044China Dept.of Atmospheric and Oceanic Sciences and Institute of Atmospheric Sciences Fudan UniversityShanghai 200438China CMA-FDU Joint Laboratory of Marine Meteorology Shanghai 200438China

Recent studies on tropical cyclone(TC)intensity change indicate that the development of a vertically aligned TC circulation is a key feature of its rapid intensification(RI),however,understanding how vortex alignment occurs remains a challenging topic in TC intensity change *** on the simulation outputs of North Atlantic Hurricane Wilma(2005)and western North Pacific Typhoon Rammasun(2014),vortex track oscillations at different vertical levels and their associated role in vortex alignment are examined to improve our understanding of the vortex alignment during RI of TCs with initial hurricane *** is found that vortex tracks at different vertical levels oscillate consistently in speed and direction during the RI of the two simulated *** the consistent track oscillation reduces the oscillation tilt during RI,the reduction of vortex tilt results mainly from the mean track before *** is also found that the vortex tilt is primarily due to the mean vortex track before and after *** track oscillations are closely associated with wavenumber-1 vortex Rossby waves that are dominant wavenumber-1 circulations in the TC inner-core *** study suggests that the dynamics of the wavenumber-1 vortex Rossby waves play an important role in the regulation of the physical processes associated with the track oscillation and vertical alignment of TCs.

关键词： tropical cyclone rapid intensification vortex tilt Rossby wave