检索结果-南通市图书馆

Rts: learning robustly from time series data with noisy label

在线全文

学校读者我要写书评

暂无评论

Frontiers of Computer Science 2024年第6期18卷 186332-186332页

作者： ZHOU Zhi JIN Yi-Xuan Li Yu-Feng national Key Laboratory for Novel Software Technology Nanjing university

Significant progress has been made in machine learning with large amounts of clean labels and static data. However, in many real-world applications, the data often changes with time and it is difficult to obtain massive clean annotations, that is, noisy labels and time series are faced simultaneously. For example, in product-buyer evaluation, each sample records the daily time behavior of users, but the long transaction period brings difficulties to analysis, and salespeople often erroneously annotate the user’s purchase behavior. Such a novel setting, to our best knowledge, has not been thoroughly studied yet, and there is still a lack of effective machine learning methods. In this paper, we present a systematic approach RTS both theoretically and empirically, consisting of two components, Noise-Tolerant Time Series Representation and Purified Oversampling Learning. Specifically, we propose reducing label noise’s destructive impact to obtain robust feature representations and potential clean samples. Then, a novel learning method based on the purified data and time series oversampling is adopted to train an unbiased model. Theoretical analysis proves that our proposal can improve the quality of the noisy data set. Empirical experiments on diverse tasks, such as the house-buyer evaluation task from real-world applications and various benchmark tasks, clearly demonstrate that our new algorithm robustly outperforms many competitive methods.

关键词： weakly-supervised learning time-series classification class-imbalanced learning

Mo_(2)B_(2)O_(2)MBene for Efficient Electrochemical CO Reduction to C_(2)Chemicals:Computational Exploration

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Energy & Environmental Materials 2024年第6期7卷 342-350页

作者： Bikun Zhang Jianwen Jiang Department of Chemical and Biomolecular Engineering National University of SingaporeSingapore 117576Singapore

Emerging as a new class of two-dimensional materials with atomically thin layers,MBenes have great potential for many important applications such as energy storage and *** mitigating carbon footprint,there has been increasing interest in CO_(2)/CO conversion on MBenes,but mostly focused on C_(1)products.C^(2+)chemicals generally possess higher energy densities and wider applications than C_(1)***,C–C coupling is technically challenging because of high energy requirement and currently few catalysts are suited for this ***,we explore electrochemical CO reduction reaction to C_(2)chemicals on Mo_(2)B_(2)O_(2)MBene via density-functional theory ***,the most favorable CO–COH coupling is revealed to be a spontaneous and barrierless process,making Mo_(2)B_(2)O_(2)an efficient catalyst for C–C *** C_(1)and C_(2)chemicals,ethanol is predicted to be the primary ***,by charge and bond analysis,it is unraveled that there exist significantly more unbonded electrons in the C atom of intermediate*COH than other C_(1)intermediates,which is responsible for the facile C–C *** an atomic scale,this work provides microscopic insight into C–C coupling process and suggests Mo_(2)B_(2)O_(2)a promising catalyst for electrochemical CO reduction to C_(2)chemicals.

关键词： C_(2)chemicals C-C coupling density-functional theory MBene Mo2B2O2

Effect of Plastic Deformation on Microstructure and Properties of Cu-(1 wt%-6 wt%) Ag Alloy

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Journal of Wuhan university of Technology(Materials Science) 2024年第3期39卷 747-753页

作者：茹亚东 ZHANG Zhongyuan 高召顺 ZHANG Ling ZUO Tingting XUE Jiangli TANG Zhixiang DA Bo LIU Yongsheng XIAO Liye Institute of Electrical Engineering Chinese Academy of SciencesBeijing 100190China Department of Physics Shanghai University of Electric PowerShanghai 201306China university of Chinese Academy of Sciences Beijing 100190China International Joint Labora-tory for Light Alloys(MOE) College of Materials Science and EngineeringChongqing UniversityChongqing 400044China Research and Services Division of Materials Data and Integrated System National Institute for Materials ScienceTsukuba 305-0047Japan

In the present study,the Cu-(1 wt%-6 wt%)Ag alloys were prepared by melting,forging and wire *** effects of plastic deformation on microstructure evolution and properties of the alloys were *** results show that non-equilibrium eutectic colonies exist in the Cu-(3 wt%-6 wt%)Ag alloy and no eutectic colonies in the 1 wt%-2 wt%Ag containing *** eutectic colonies are aligned along the drawing direction and refined with the increase of draw *** to the refinement of eutectic colonies,the Cu-Ag alloy exhibits higher strength with the increase of draw *** Cu-6Ag alloy exhibits excellent comprehensive properties with a strength of 930 MPa and a conductivity of 82%IACS when the draw ratio reaches 5.7.

关键词： Cu-Ag alloy high strength and high conductivity microstructure eutectic structure strengthening mechanism

GS-orthogonalization OMP method for space target detection via bistatic space-based radar

在线全文

学校读者我要写书评

暂无评论

Chinese Journal of Aeronautics 2024年第7期37卷 333-351页

作者： Shuyu ZHENG Libing JIANG Qingwei YANG Yingjian ZHAO Zhuang WANG national Key Laboratory of Science and Technology on Automatic Target Recognition College of Electronic Science and TechnologyNational University of Defense TechnologyChangsha 410073China

A space-based bistatic radar system composed of two space-based radars as the transmitter and the receiver respectively has a wider surveillance region and a better early warning capability for high-speed targets,and it can detect focused space targets more flexibly than the monostatic radar system or the ground-based radar ***,the target echo signal is more difficult to process due to the high-speed motion of both space-based radars and space *** be specific,it will encounter the problems of Range Cell Migration(RCM)and Doppler Frequency Migration(DFM),which degrade the long-time coherent integration performance for target detection and localization *** solve this problem,a novel target detection method based on an improved Gram Schmidt(GS)-orthogonalization Orthogonal Matching Pursuit(OMP)algorithm is proposed in this ***,the echo model for bistatic space-based radar is constructed and the conditions for RCM and DFM are ***,the proposed GS-orthogonalization OMP method is applied to estimate the equivalent motion parameters of space ***,the RCM and DFM are corrected by the compensation function correlated with the estimated motion ***,coherent integration can be achieved by performing the Fast Fourier Transform(FFT)operation along the slow time direction on compensated echo *** simulations and real raw data results validate that the proposed GS-orthogonalization OMP algorithm achieves better motion parameter estimation performance and higher detection probability for space targets detection.

关键词： Bistatic space-based radar High-speed maneuvering space targets detection Range Cell Migration(RCM) Doppler Frequency Migration(DFM) Gram Schmidt(GS)-orthogonalization Orthogonal Matching Pursuit(OMP)algorithm

Advances of Pipeline Model Parallelism for Deep Learning Training:An Overview

在线全文

学校读者我要写书评

暂无评论

Journal of Computer Science & Technology 2024年第3期39卷 567-584页

作者：关磊李东升梁吉业王文剑葛可适卢锡城 College of Science National University of Defense TechnologyChangsha 410073China College of Computer National University of Defense TechnologyChangsha 410073China School of Computer and Information Technology Shanxi UniversityTaiyuan 030006China CCF IEEE

Deep learning has become the cornerstone of artificial intelligence,playing an increasingly important role in human production and ***,as the complexity of problem-solving increases,deep learning models become increasingly intricate,resulting in a proliferation of large language models with an astonishing number of *** model parallelism(PMP)has emerged as one of the mainstream approaches to addressing the significant challenge of training“big models”.This paper presents a comprehensive review of *** covers the basic concepts and main challenges of *** also comprehensively compares synchronous and asynchronous pipeline schedules for PMP approaches,and discusses the main techniques to achieve load balance for both intra-node and inter-node ***,the main techniques to optimize computation,storage,and communication are presented,with potential research directions being discussed.

关键词： deep learning pipeline schedule load balance multi-GPU system pipeline model parallelism(PMP)

Surface engineering with ionic polymers on membranes for boron removal

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Frontiers of Chemical Science and Engineering 2024年第5期18卷 61-69页

作者： Xiting Zhang Chenyi Fang J Paul Chen Sui Zhang Department of Chemical and Biomolecular Engineering National University of SingaporeSingapore 117576Singapore Department of Civil and Environmental Engineering National University of SingaporeSingapore 119260Singapore College of Chemistry and Environmental Engineering Shenzhen UniversityShenzhen 518060China

Removal of boric acid from seawater and wastewater using reverse osmosis membrane technologies is imperative and yet remains inadequately addressed by current commercial *** research efforts performed post-modification of reverse osmosis membranes to enhance boron rejection,which is usually accompanied by substantial sacrifice in water *** study delves into the surface engineering of low-pressure reverse osmosis membranes,aiming to elevate boron removal efficiency while maintaining optimal salt rejection and water *** were modified by the self-polymerization and co-deposition of dopamine and polystyrene sulfonate at varying ratios and *** surfaces became smoother and more hydrophilic after *** optimum membrane exhibited a water permeability of 9.2±0.1 L·m^(-2)·h^(-1)·bar^(-1),NaCl rejection of 95.8%±0.3%,and boron rejection of 49.7%±0.1% and 99.6%±0.3% at neutral and alkaline pH,*** water permeability is reduced by less than 15%,while the boron rejection is 3.7 times higher compared to the blank *** research provides a promising avenue for enhancing boron removal in reverse osmosis membranes and addressing water quality concerns in the desalination process.

关键词： membrane low-pressure reverse osmosis boron surface engineering

Improved Simulation of Summer Heavy Rainfall over Beijing and Henan by the YHGSM with Updated Subgrid Orographic Parameters

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Journal of Meteorological Research 2024年第3期38卷 504-529页

作者： Yingjie WANG Jianping WU Kaijun REN College of Computer Science and Technology National University of Defense TechnologyChangsha410600China College of Meteorology and Oceanography National University of Defense TechnologyChangsha410600China

In numerical weather prediction(NWP),the parameterization of orographic drag plays an important role in representing subgrid orographic *** subgrid orographic parameters are the key input to the parameterization of orographic ***,the subgrid orographic parameters in most NWP models were produced based on elevation datasets generated many years ago,with a coarse resolution and low *** this paper,using the latest high-quality elevation data and considering the applicable scale range of the subgrid orographic parameters,we construct the orographic parameters,including the subgrid orographic standard deviation,anisotropy,orientation,and slope,that are required as input to the orographic gravity wave drag(OGWD)***,we introduce the newly constructed orographic parameters into the Yin-He Global Spectral Model(YHGSM),optimize the description of the orographic effect in the model,and improve the simulation of two typical heavy rainfall events in Beijing and Henan.

关键词： subgrid orographic parameters orographic drag heavy rainfall Yin-He Global Spectral Model(YHGSM) parameterization

Efficient single‑pixel imaging based on a compact fiber laser array and untrained neural network

在线全文

学校读者我要写书评

暂无评论

Frontiers of Optoelectronics 2024年第1期17卷 85-93页

作者： Wenchang Lai Guozhong Lei Qi Meng Yan Wang Yanxing Ma Hao Liu Wenda Cui Kai Han College of Advanced Interdisciplinary Studies National University of Defense TechnologyChangsha 410073China Nanhu Laser Laboratory National University of Defense TechnologyChangsha 410073China

This paper presents an efficient scheme for single-pixel imaging(SPI)utilizing a phase-controlled fiber laser array and an untrained deep neural *** fiber lasers are arranged in a compact hexagonal structure and coherently combined to generate illuminating light *** the utilization of high-speed electro-optic modulators in each individual fiber laser module,the randomly modulated fiber laser array enables rapid speckle projection onto the object of ***,the untrained deep neural network is incorporated into the image reconstructing process to enhance the quality of the reconstructed *** simulations and experiments,we validate the feasibility of the proposed method and successfully achieve high-quality SPI utilizing the coherent fiber laser array at a sampling ratio of 1.6%.Given its potential for high emitting power and rapid modulation,the SPI scheme based on the fiber laser array holds promise for broad applications in remote sensing and other applicable fields.

关键词： Single-pixel imaging Fiber laser array Deep learning Remote sensing

Cross-Domain Bilateral Access Control on Blockchain-Cloud Based Data Trading System

在线全文

学校读者我要写书评

暂无评论

Computer Modeling in Engineering & Sciences 2024年第10期141卷 671-688页

作者： Youngho Park Su Jin Shin Sang Uk Shin Electronics and Information Communications Research Center Pukyong National UniversityBusan48513Republic of Korea Department of Information Security Graduate SchoolPukyong National UniversityBusan48513Republic of Korea Division of Computer Engineering Pukyong National UniversityBusan48513Republic of Korea

Data trading enables data owners and data requesters to sell and purchase *** the emergence of blockchain technology,research on blockchain-based data trading systems is receiving a lot of ***,to reduce the on-chain storage cost,a novel paradigm of blockchain and cloud fusion has been widely considered as a promising data trading ***,the fact that data can be used for commercial purposes will encourage users and organizations from various fields to participate in the data *** the data marketplace,it is a challenge how to trade the data securely outsourced to the external cloud in a way that restricts access to the data only to authorized users across multiple *** this paper,we propose a cross-domain bilateral access control protocol for blockchain-cloud based data trading *** consider a system model that consists of domain authorities,data senders,data receivers,a blockchain layer,and a cloud *** proposed protocol enables access control and source identification of the outsourced data by leveraging identity-based cryptographic *** the proposed protocol,the outsourced data of the sender is encrypted under the target receiver’s identity,and the cloud provider performs policy-match verification on the authorization tags of the sender and receiver generated by the identity-based signature ***,data trading can be achieved only if the identities of the data sender and receiver simultaneously meet the policies specified by each *** demonstrate efficiency,we evaluate the performance of the proposed protocol and compare it with existing studies.

关键词： Bilateral access control blockchain data sharing policy-match