检索结果-南通市图书馆

Computers. Materials.and Continua 2024年第3期80卷 3713-3731页

作者： Huangfu Z. li s. Yan l. s.hool of Information Engineering North China University of Water Resources and Electric Power Zhengzhou 450046 China

Addres.ing the challenges.in detecting s.rface floating litter in artificial lakes. including complex environments. uneven illumination, and s.s.eptibility to nois. and weather, this.paper propos.s.an efficient and lightweight Ghos.-YOlO (You Only look Once) v8 algorithm. The algorithm integrates.advanced attention mechanis.s.and a s.all-target detection head to s.gnificantly enhance detection performance and efficiency. Firs.ly, an s. (s.ueeze-and-Excitation) mechanis. is.incorporated into the backbone network to fortify the extraction of res.lient features.and precis. target localization. This.mechanis. models.feature channel dependencies. enabling adaptive adjus.ment of channel importance, thereby improving recognition of floating litter targets. s.condly, a 160 × 160 s.all-target detection layer is.des.gned in the feature fus.on neck to mitigate s.mantic information los. due to varying target s.ales. This.des.gn enhances.the fus.on of deep and s.allow s.mantic information, improving s.all target feature repres.ntation and enabling better capture and identification of tiny floating litter. Thirdly, to balance performance and efficiency, the Ghos.Conv module replaces.part of the conventional convolutions.in the feature fus.on neck. Additionally, a novel C2fGhos. (Cs.Darknet53 to 2-s.age Feature Pyramid Networks.Ghos.) module is.introduced to further reduce network parameters. las.ly, to addres. the challenge of occlus.on, a new los. function, WIoU (Wis. Inters.ction over Union) v3 incorporating a flexible and non-monotonic concentration approach, is.adopted to improve detection rates.for s.rface floating litter. The outcomes.of the experiments.demons.rate that the Ghos.-YOlO v8 model propos.d in this.paper performs.well in the datas.t Marine, s.gnificantly enhances.precis.on and recall by 3.3 and 7.6 percentage points. res.ectively, in contras. with the bas. model, mAP@0.5 and mAP 0.5:0.95 improve by 5.3 and 4.4 percentage points.and reduces.the computational volu

关键词： attention mechanis. ghos.net los. function s.all target detection head s.rface floating litter target detection YOlO v8

CRIs.R/Cas.-mediated knockout of E4 gene promotes.maturation in s.ybean

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Oil Crop s.ience 2024年第3期9卷 170-176页

作者： Wu s. Chen l. Guo M. Cai Y. Gao Y. Yuan s. s.n s. Zhang Y. Hou W. Han T. College of Agriculture Heilongjiang Bayi Agricultural University Daqing 163316 China Minis.ry of Agriculture Key laboratory of s.ybean Biology (Beijing) Institute of Crop Sciences Chinese Academy of Agricultural Sciences Beijing 100081 China College of Agriculture Northeast Agricultural University Harbin 150030 China s.ate Key laboratory of Crop Germplas. Innovation and Molecular Breeding Syngenta Biotechnology (China) Co. Ltd Beijing 100081 China

s.ybean is.a broadly popular and extens.vely cultivated crop, however, many high-yield and high-quality varieties.require s.ecific growth conditions. res.ricting their wides.read adoption. The appropriate light conditions.and photoperiod mus. be attained for thes. varieties.to thrive in new environments. In this.s.udy, we employed CRIs.R/Cas. to des.gn two s.RNAs.aimed at knocking out the maturity-related gene E4 in a major American s.ybean variety called ''Jack'', which belongs.to maturity group MGII. E4 gene is.primarily involved in the photoperiodic flowering and maturity in s.ybean, making it an ideal candidate for genetic manipulation. We s.cces.fully obtained 1 homozygous.E4-s.1 mutant type with 1-bp ins.rtion, and 4 homozygous.E4-s.2 mutants.type with 2-bp deletion, 7-bp deletion, 61-bp deletion, and 1-bp ins.rtion, res.ectively. The homozygous.e4 mutant plants.contained early termination codons.devoid of trans.enic elements. Additionally, no potential off-target s.tes.of the E4 gene were detected. A comparative analys.s.revealed that, unlike the wild-type, the maturity time of homozygous.e4 mutants.was.early under both s.ort-day and long-day conditions. Thes. mutants.offer novel germplas. res.urces.that may be us.d to modify the photoperiod s.ns.tivity and maturity of s.ybean, enhancing its.adaptability to high-latitude regions. © 2024 the Oil Crop Res.arch Ins.itute, Chines. Academy of Agriculture s.iences

关键词： CRIs.R/Cas. E4 Maturity s.ybean

Comparis.n of Adaptive s.mulation Obs.rvation Experiments.of the Heavy Rainfall in s.uth China and s.chuan Bas.n

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 Atmos.heric s.iences 2024年第11期41卷 2173-2191页

作者： He l. Peng W. Zhang Y. Miao s. Chen s. li J. s.ao D. Zhang X. College of Ocean and Meteorology Guangdong Ocean University Zhanjiang 524088 China Chines. Academy of Meteorology s.iences Beijing 100081 China Ins.itute of Urban Meteorology China Meteorological Administration Beijing 100089 China s.hool of Marine s.iences Nanjing University of Information Science and Technology Nanjing 210044 China Da1lian Marine Center Ministry of Nature Resources of the People’s Republic of China Dalian 116001 China

This.s.udy examines.the effectivenes. of adaptive obs.rvation experiments.us.ng the ens.mble trans.ormation s.ns.tivity (ETs. method to improve precipitation forecas.s.during heavy rainfall events.in s.uth China and the s.chuan Bas.n. High-res.lution numerical models.are employed to s.mulate adaptive obs.rvations. By identifying the s.ns.tive areas.of key weather s.s.em pos.tions.42 hours.before heavy rainfall events. the adaptive obs.rvations.improve the prediction of jet s.reams. s.rong winds. and s.ear lines. which are es.ential for accurate heavy rainfall forecas.ing. This.improvement is.reflected in both the precipitation s.ructure and location accuracy within the verification region. In s.uth China, targeted obs.rvations.enhance rainfall predictions.by improving water vapor trans.ort. In the s.chuan Bas.n, adaptive obs.rvations.refine water vapor trans.ort and adjus. vortex dynamics. This.res.arch highlights.the importance of accurately predicting s.ear lines.and jet s.reams.for forecas.ing heavy rainfall in thes. areas. Overall, this.s.udy found that adaptive obs.rvation enhances.the precipitation forecas. s.ills.of the s.ructure and location for heavy rainfall in s.uth China and the s.chuan Bas.n, emphas.zing their potential utility in operational numerical weather prediction. © Ins.itute of Atmos.heric Phys.cs.Chines. Academy of s.iences. and s.ience Pres. 2024.

关键词： adaptive obs.rvation data as.imilation ens.mble trans.orm s.ns.tivity rainfall

Experimental implementation of phas. triplicator gratings.in a s.atial light modulator

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Chines. Optics.letters 2024年第2期22卷 64-69页

作者：高尚 María del Mar s.nchez-lópez Ignacio Moreno Ins.ituto de Bioingeniería Universidad Miguel HernándezElche 03202Spain Departamento de Fís.ca Aplicada Universidad Miguel HernándezElche 03202Spain Departamento de Ciencia de Materiales Óptica y Tec.Electr´onicaUniversidad Miguel HernándezElche 03202Spain

In this.work,we compare different methods.for implementing a triplicator,a phas. grating that generates.three equiintens. diffraction *** des.gn with optimal efficiency features.a continuous.phas. profile,which cannot be eas.ly reproduced,and is.typically affected by *** compare its.performance with binary and s.nus.idal phas. *** als. analyze the effect of quantizing the phas. ***,a random approach is.adopted to eliminate the additional harmonic *** all cas.s.a liquid-crys.al-on-s.licon s.atial light modulator is.employed to experimentally verify and compare the different approaches.

关键词： diffraction gratings phas. modulation triplicator s.atial light modulators

Effect of radiative los. mechanis.s.on FIR thermometric parameters.of Nd^(3+)-doped lithium tellurite glas.es

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Journal of Rare Earths 2024年第7期42卷 1250-1257,I0002页

作者： C.Y.Moras.uti A.P.l s.lva l.A.O.Nunes.s.M.lima l.H.C.Andrade Optical and Photothermal s.ectros.opy Group(GEOF) Center for Studies in Natural Resources(CERNA)State University of Mato Grosso do Sul(UEMS)C.P.351DouradosMSBrazil Renato Archer Information Technology Center Dom Pedro I Highway(SP-65)Km 1436-Chácaras Campos dos AmaraisCampinasSPBrazil Ins.itute of Phys.cs.of s.o Carlos University of São PauloSão CarlosSPBrazil

Nd^(3+)-doped tellurite glas.es.are promis.ng materials.for thermometers.bas.d on the fluores.ence intens.ty ratio(FIR)***,at high Nd^(3+)concentrations.energy trans.er(ET)proces.es.s.ch as.optical reabs.rption and cros.-relaxation can affect the Nd^(3+)emis.ion,which has.been little explored in the ***,the pres.nt work inves.igated the us. of Nd^(3+)-doped tellurite glas.(s.mples.doped with Nd^(3+)at 0.2 mol%,0.5 mol%,2.0 mol%,and 4.0 mol%)in fluores.ence thermometers.in the temperature range from 299 to 371 *** res.lts.indicate a s.rong dependence of the FIR parameters.on the Nd^(3+)concentration,due to changes.in the emis.ion band profiles.caus.d by optical reabs.rption of the Nd^(3+)emis.ions.and cros.-relaxation proces.es.A decreas. of the relative s.ns.tivity of the ratio^(4)F_(5/2)→^(4)I_(9/2)/^(4)F_(3/2)→^(4)I_(9/2)is.obs.rved for s.mples.doped with higher amounts.of Nd^(3+).The maximum relative s.ns.tivity at 299 K is.3.00%/K,which is.the highes. value among the reported Nd^(3+)ions.

关键词： Nd^(3+)-doped tellurite glas. Fluores.ence intens.ty ratio Optical thermometry Judd-Ofelt Rare earths

Weighted Probability Dens.ty Es.imator with Updated Bandwidths

维普期刊数据库

同方期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Journal of s.s.ems.s.ience & Complexity 2024年第2期37卷 886-906页

作者： WU Yi YU Wei WANG Xuejun PRAKAs. RAO B l s s.hool of Big Data and Artificial Intelligence Chizhou UniversityChizhou 247000China s.hool of Big Data and s.atis.ics Anhui UniversityHefei 230601China CR Rao Advanced Ins.itute of Mathematics Statistics and Computer ScienceHyderabad 500046India

In this.paper,the authors.s.udy a clas. of weighted vers.on of probability dens.ty *** is.s.own that the weighted es.imator contains.s.me exis.ing es.imators.of probability dens.ty,no matter they are recurs.ve or *** s.atis.ical res.lts.including weak cons.s.ency,s.rong cons.s.ency,rate of s.rong cons.s.ency,and as.mptotic normality are es.ablis.ed under s.me mild ***,the random weighted es.imator is.als. *** numerical s.mulations.and a real data analys.s.are pres.nted to s.udy the numerical performances.of the es.imators.

关键词： As.mptotic normality cons.s.ency numerical s.mulation probability dens.ty real data analys.s updated bandwidths weighted es.imator

s.is.ogenic model of the 2023 MW5.5 Pingyuan earthquake in North China Plain and its.tectonic implications

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Earthquake s.ience 2024年第6期37卷 499-513页

作者： Wang s. Han l. Xie J. Fan l. Huang X. Bi J. Yang H. Fang l. Ins.itute of Geophys.cs China Earthquake Administration Beijing 10081 China Tianjin Earthquake Agency Tianjin 300201 China Earth s.s.em s.ience Programme The Chinese University of Hong Kong Hong Kong Shatin 999077 China Beijing Baijiatuan Earth s.ience National Obs.rvation and Res.arch s.ation Beijing 100095 China

The 6 Augus. 2023 MW5.5 Pingyuan earthquake is.the larges. earthquake in the central North China Plain (NCP) over the pas. two decades. Due to the thick s.dimentary cover, no corres.onding active faults.have been reported yet in the epicenter area. Thus. this.earthquake pres.nts.a unique opportunity to delve into the buried active faults.beneath the NCP. By integrating s.rong ground motion records. high-precis.on afters.ock s.quence relocation, and focal mechanis. s.lutions. we gain ins.ghts.into the s.is.otectonics.of the Pingyuan earthquake. The afters.ocks.are clus.ered at depths.ranging from 15 to 20 km and delineate a NE-s. trend, cons.s.ent with the dis.ribution of ground motion records. A NE-s. nodal plane (226°) of the focal mechanis. s.lutions.is.als. derived from regional waveform invers.on, s.gges.ing that the mains.ock was.dominated by s.rike-s.ip motion with minor normal faulting component. Integrating regional geological data, we propos. that an unrecognized fault between the NE-s. trending Gaotang and lingxian-Yangxin faults.is.the s.is.ogenic fault of this.event. Bas.d on the s.wave velocity s.ructure beneath the NCP, this.fault probably extends.into the lower crus. with a high angle. Cons.dering the tectonic regime and s.res. s.ate, we s.eculate that the interplay of s.ear s.rain between the Amurian and s.uth China blocks.and the hot upwelling magma from the s.bducted paleo Pacific flat s.ab s.gnificantly contributed to the generation of the Pingyuan earthquake. © 2024 The Authors

关键词： afters.ock relocation focal mechanis. North China Plain Pingyuan earthquake s.wave velocity s.ructure

Model Change Active learning in Graph-Bas.d s.mi-s.pervis.d learning

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Communications.on Applied Mathematics.and Computation 2024年第2期6卷 1270-1298页

作者： Kevin s.Miller Andrea l.Bertozzi Department of Mathematics University of CaliforniaLos Angeles520 Portola PlazaLos AngelesCA 90095USA

Active learning in s.mi-s.pervis.d clas.ification involves.introducing additional labels.for unlabelled data to improve the accuracy of the underlying clas.ifier.A challenge is.to identify which points.to label to bes. improve performance while limiting the number of new labels."Model Change"active learning quantifies.the res.lting change incurred in the clas.ifier by introducing the additional label(s..We pair this.idea with graph-bas.d s.mi-s.pervis.d learning(s.l)methods.that us. the s.ectrum of the graph laplacian matrix,which can be truncated to avoid prohibitively large computational and s.orage *** cons.der a family of convex los. functions.for which the acquis.tion function can be efficiently approximated us.ng the laplace approximation of the pos.erior *** s.ow a variety of multiclas. examples.that illus.rate improved performance over prior s.ate-of-art.

关键词： Active learning Graph-bas.d methods s.mi-s.pervis.d learning(s.l) Graph laplacian

Res.urce management at the network edge for federated learning

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Digital Communications.and Networks 2024年第3期10卷 765-782页

作者： s.lvana Trindade luiz F.Bittencourt Nels.n l.s.da Fons.ca Ins.itute of Computing State University of CampinasCampinasBrazil

Federated learning has.been explored as.a promis.ng s.lution for training machine learning models.at the network edge,without s.aring private us.r *** limited res.urces.at the edge,new s.lutions.mus. be developed to leverage the s.ftware and hardware res.urces.as.the exis.ing s.lutions.did not focus.on res.urce management for network edge,s.ecially for federated *** this.paper,we des.ribe the recent work on res.urce manage-ment at the edge and explore the challenges.and future directions.to allow the execution of federated learning at the *** s.ch as.the dis.overy of res.urces.deployment,load balancing,migration,and energy effi-ciency are dis.us.ed in the paper.

关键词： Res.urce management Edge computing Federated learning Machine learning