检索结果-南通市图书馆

Research on three-step accelerated gradient algorithm in deep learning

在线全文

学校读者我要写书评

暂无评论

Statistical Theory and Related Fields 2022年第1期6卷 40-57页

作者： yongqiang lian Yincai Tang Shirong Zhou KLATASDS-MOE School of StatisticsEast China Normal UniversityShanghaiPeople's Republic of China

Gradient descent(GD)algorithm is the widely used optimisation method in training machine learning and deep learning *** this paper,based on GD,Polyak’s momentum(PM),and Nesterov accelerated gradient(NAG),we give the convergence of the algorithms from an ini-tial value to the optimal value of an objective function in simple quadratic *** on the convergence property of the quadratic function,two sister sequences of NAG’s iteration and par-allel tangent methods in neural networks,the three-step accelerated gradient(TAG)algorithm is proposed,which has three sequences other than two sister *** illustrate the perfor-mance of this algorithm,we compare the proposed algorithm with the three other algorithms in quadratic function,high-dimensional quadratic functions,and nonquadratic *** we consider to combine the TAG algorithm to the backpropagation algorithm and the stochastic gradient descent algorithm in deep *** conveniently facilitate the proposed algorithms,we rewite the R package‘neuralnet’and extend it to‘supneuralnet’.All kinds of deep learning algorithms in this paper are included in‘supneuralnet’***,we show our algorithms are superior to other algorithms in four case studies.

关键词： Accelerated algorithm backpropagation deep learning learning rate momentum stochastic gradient descent

Revisiting thermal comfort and thermal sensation

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Building Simulation 2024年第2期17卷 185-188页

作者： Zhiwei lian Department of Architecture Shanghai Jiao Tong UniversityShanghai200240China

The thermal environment is closely intertwined with comfort,health,efficiency,and building energy *** Comfort stands as a crucial concept for evaluating the thermal environment,yet the prevailing evaluation metric is Thermal *** correspondence between genuine Thermal Comfort and Thermal Sensation is influenced by a variety of environmental factors.

关键词： thermal Comfort prevailing

Modeling and Simulation of Lithium Vacuum Evaporation Process Using COMSOL Multiphysics

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 Thermal Science 2024年第1期33卷 86-100页

作者： lian Xuexin ZHONG Dawen Beijing Key Laboratory of Passive Nuclear Power Safety and Technology North China Electric Power UniversityBeijing 102206China

Based on the COMSOL Multiphysics simulation software,this study carried out modeling and numerical simulation for the evaporation process of liquid metal lithium in the vacuum free molecular flow *** motion of lithium atoms in the evaporation process was analyzed through a succession of *** on the available experimental values of the saturated vapor pressure of liquid metal lithium,the relationship between saturated vapor pressure and temperature of liquid lithium in the range of 600 K-900 K was obtained.A two-dimensional symmetric model(3.5 mm×20 mm) was established to simulate the transient evaporation process of liquid lithium at wall temperatures of 750 K,780 K,800 K,810 K,825 K,and 850 K,*** effects of temperature,the evaporation coefficient,back pressure,and length-to-diameter ratio on the evaporation process were studied;the variation trends and reasons of the molecular flux and the pressure during the evaporation process were *** the same time,the evaporation process under variable wall temperature conditions was *** research made the evaporation process of liquid lithium in vacuum molecular flow clearer,and provided theoretical support for the space reactor and nuclear fusion related fields.

关键词： free molecular flow saturated vapor pressure vacuum evaporation liquid lithium metal

DeepSI:A Sensitive-Driven Testing Samples Generation Method of Whitebox CNN Model for Edge Computing

在线全文

学校读者我要写书评

暂无评论

Tsinghua Science and Technology 2024年第3期29卷 784-794页

作者： Zhichao lian Fengjun Tian School of Cyber Science and Engineering Nanjing University of Science and TechnologyNanjing 210094China

In recent years,Deep Learning(DL)technique has been widely used in Internet of Things(IoT)and Industrial Internet of Things(IIoT)for edge computing,and achieved good *** more and more studies have shown the vulnerability of neural ***,it is important to test the robustness and vulnerability of neural *** specifically,inspired by layer-wise relevance propagation and neural network verification,we propose a novel measurement of sensitive neurons and important neurons,and propose a novel neuron coverage criterion for robustness *** on the novel criterion,we design a novel testing sample generation method,named DeepSI,which involves definitions of sensitive neurons and important ***,we construct sensitive-decision paths of the neural network through selecting sensitive neurons and important ***,we verify our idea by setting up several experiments,then results show our proposed method achieves superior performances.

关键词： neuron sensitivity Layer-wise Relevance Propagation(LRP) neural network verification deeplearning testing

Contrasting physical mechanisms of yellowfin tuna fluctuations between the western and eastern Indian Ocean

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 Oceanology and Limnology 2024年第3期42卷 960-971页

作者： Peng lian Le GAO Key Laboratory of Ocean Observation and Forecasting and Key Laboratory of Ocean Circulation and Waves Institute of OceanologyChinese Academy of SciencesQingdao 266071China University of Chinese Academy of Sciences Beijing 100049China Laboratory for Ocean Dynamics and Climate Laoshan LaboratoryQingdao 266061China

As an economically critical pelagic migratory species,yellowfin tuna(Thunnus albacores,YFT)is very sensible to physical and environmental conditions,such as sea surface temperature(SST),ocean heat content(OHC),and the mixed layer depth(MLD).We investigated the impact of SST,OHC,and MLD on fluctuations of YFT catch in the western/eastern Indian Ocean using the long time series of 63-year environmental and YFT *** found that the impact of SST on YFT was heavily overestimated in the past,and MLD plays a more critical role in the YFT catch *** the MLD deepens(>34.8 m),SST was more influential in predicting the catches of YFT than OHC in the western Indian Ocean,and OHC was more critical to YFT than SST in the eastern Indian ***,when the MLD shallows(<34.8 m),MLD was more vital to predict the catch per unit effort(CPUE)of YFT than SST/OHC in the *** 2000,there was an asynchronous pattern of YFT CPUE induced by higher frequency variations and ocean hiatus of SST/OHC signals in the western and eastern Indian Oceans *** impact of the subsurface hiatus may induce the decrease of YFT in the eastern Indian *** above findings clarified a non-stationary relationship between the environmental factors and catches of YFT and provided new insights into variations in YFT abundance.

关键词： yellowfin tuna Indian Ocean sea surface temperature(SST) ocean heat content(OHC) mixed layer depth(MLD) hiatus

Impacts of central-Pacific El Niño and physical drivers on eastern Pacific bigeye tuna

在线全文

学校读者我要写书评

暂无评论

Journal of Oceanology and Limnology 2024年第3期42卷 972-987页

Bigeye tuna Thunnus obesus is an important migratory species that forages deeply,and El Niño events highly influence its distribution in the eastern Pacific *** sea surface temperature is widely recognized as the main factor affecting bigeye tuna(BET)distribution during El Niño events,the roles of different types of El Niño and subsurface oceanic signals,such as ocean heat content and mixed layer depth,remain *** conducted A spatial-temporal analysis to investigate the relationship among BET distribution,El Niño events,and the underlying oceanic signals to address this knowledge *** used monthly purse seine fisheries data of BET in the eastern tropical Pacific Ocean(ETPO)from 1994 to 2012 and extracted the central-Pacific El Niño(CPEN)indices based on Niño 3 and Niño ***,we employed Explainable Artificial Intelligence(XAI)models to identify the main patterns and feature importance of the six environmental variables and used information flow analysis to determine the causality between the selected factors and BET ***,we analyzed Argo datasets to calculate the vertical,horizontal,and zonal mean temperature differences during CPEN and normal years to clarify the oceanic thermodynamic structure differences between the two types of *** findings reveal that BET distribution during the CPEN years is mainly driven by advection feedback of subsurface warmer thermal signals and vertically warmer habitats in the CPEN domain area,especially in high-yield fishing *** high frequency of CPEN events will likely lead to the westward shift of fisheries centers.

关键词： bigeye tuna central-Pacific El Niño(CPEN) ocean heat content Argo explainable artificial intelligence(XAI) information flow

Emerging materials and transistors for integrated circuits

在线全文

学校读者我要写书评

暂无评论

National Science Review 2024年第3期11卷 7-8页

作者： Ming Liu lian-Mao Peng Lab of Nanofabrication and Novel Device Integration Institute of Microelectronics Chinese Academy of Sciences Key Laboratory for the Physics and Chemistry of Nanodevices and Center for Carbon-based Electronics School of Electronics Peking University

The device technology for integrated circuits(ICs) has received a great deal of attention in the past decade since the mainstream silicon-based complementary metal-oxide-semiconductor(CMOS) field-effect transistor(FET) for modern ICs had been encountering development bottlenecks due to the combination of power constraints, physics, cost and fabrication processes.

关键词：

Recent Advances in Nickel Catalyzed Carbonylative Reactions via the Insertion of Carbon Monoxide

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Chinese Journal of Chemistry 2024年第2期42卷 177-189页

作者： Xinzhou Chen Gang Chen Zhong lian Department of Dermatology State Key Laboratory of Biotherapy and Cancer CenterWest China HospitalSichuan UniversityChengduSichuan 610041China

Carbonyl compounds have attracted considerable attention due to their extensive applications in drug ***,they are important synthetic intermediates for the construction of carbon-carbon and carbon-heteroatom ***-metal-catalyzed carbonylation via the insertion of co is one of the most efficient and straightforward strategies to access carbonyl ***,most of the transition-metal-catalyzed carbonylative reactions require expensive and toxic noble-metal ***,there is a growing demand for the exploration of nickel-catalyzed carbonylative reactions via the insertion of CO due to the earth abundance and low cost of *** with the well-established palladium-catalyzed carbonylative reactions,nickel-catalyzed analogous transformations have been relatively *** is primarily because CO strongly binds to nickel,often resulting in catalyst *** recent years,some research groups have focused on using CO surrogates or NN_(2) pincer nickel catalyst to circumvent the formation of Ni(CO)_(4).Nickel-catalyzed carbonylation has been applied in the construction of carbonyl-containing compounds,such as ketones,carboxylic acids,thioesters,acyl chloride and carboxamides.

关键词： Carbonylation Nickel catalysis Carbon monoxide C1 chemistry Carbonyl-containing compounds

Diving dynamics identification and motion prediction for marine crafts using field data

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 Ocean Engineering and Science 2024年第4期9卷 391-400页

作者： Yiming Zhong Caoyang Yu Yulin Bai Zheng Zeng lian lian School of Oceanography Shanghai Jiao Tong UniversityShanghai 200030China Key Laboratory of Polar Ecosystem and Climate Change Ministry of EducationShanghai Jiao Tong UniversityShanghai 200030China Shanghai Key Laboratory of Polar Life and Environment Sciences Shanghai Jiao Tong UniversityShanghai 200030China State Key Laboratory of Ocean Engineering Shanghai Jiao Tong UniversityShanghai 200240China

Ensuring accurate parameter identification and diving motion prediction of marine crafts is essential for safe navigation,optimized operational efficiency,and the advancement of marine *** this,this paper proposes an instrumental variable-based least squares(IVLS)***,aiming to balance complexity with accuracy,a decoupled diving model is constructed,incorporating nonlinear actuator characteristics,inertia coefficients,and damping ***,a discrete parameter identification matrix is designed based on this dedicated model,and then a IVLS algorithm is innovatively derived to reject measurement ***,the stability of the proposed algorithm is validated from a probabilistic point of view,providing a solid theoretical ***,performance evaluation is conducted using four depth control datasets obtained from a piston-driven profiling float in Qiandao Lake,with desired depths of 30 m,40 m,50 m,and 60 *** on the diving dynamics identification results,the IVLS algorithm consistently shows superior performance when compared to recursive weighted least squares algorithm and least squares support vector machine algorithm across all depths,as evidenced by lower average absolute error(AVGAE),root mean square error(RMSE),and maximum absolute error values and higher determination coefficient(R2).Specifically,for desired depth of 60 m,the IVLS algorithm achieved an AVGAE of 0.553 m and RMSE of 0.655 m,significantly outperforming LSSVM with AVGAE and RMSE values of 8.782 m and 11.117 m,***,the IVLS algorithm demonstrates a remarkable generalization capability with R2 values consistently above 0.95,indicating its robustness in handling varied diving dynamics.

关键词： Marine craft Parameter identification Motion prediction Instrumental variable-based least-squares algorithm Diving dynamics model