期刊文献+
共找到766篇文章
< 1 2 39 >
每页显示 20 50 100
Effects of Group Size on Vigilance Behavior of Wintering Common Cranes Grus grus 被引量:12
1
作者 杨洋 陈文华 +3 位作者 江望高 杨士剑 彭贵鸿 黄庭发 《Zoological Research》 CAS CSCD 北大核心 2006年第4期357-362,共6页
From January to March 2005, the effects of group size on the vigilance behavior of wintering Common Cranes Grus grus were studied at Lashihai Lake, Yunnan Province, China. We used scan sampling to record the group siz... From January to March 2005, the effects of group size on the vigilance behavior of wintering Common Cranes Grus grus were studied at Lashihai Lake, Yunnan Province, China. We used scan sampling to record the group sizes and the number of vigilant individuals in each group, and focal sampling to record the frequency and duration of individual vigilance behavior. Both the vigilance efforts of groups and individuals significantly decreased as group size increased, but when the group size exceeded 30 individuals, the decrease of group vigilance became not significant (P 〉 0. 05), and the vigilance duration of adult cranes increased (P 〈 0.01 ). The vigilance frequency of adults increased (P 〈 0.05) when the size exceeded 50 individuals. Presumably, the maximal group size allowing the lowest vigilance for juvenile cranes was larger than that for adults, and the flocks composed of 20 to 30 individuals represented the optimal group size of wintering Common Cranes by considering only the vigilance behavior. Further research should focus on the synthesized effects of various factors. 展开更多
关键词 grus grus Group size Vigilance effort Frequency DURATION
在线阅读 下载PDF
Genetic structure of wintering Hooded Cranes (Grus monacha) based on mitochondrial DNA D-loop sequences 被引量:3
2
作者 张黎黎 周立志 代艳丽 《Chinese Birds》 2012年第2期71-81,共11页
The Hooded Crane (Grus monacha) is a waterbird wintering in the wetlands of the middle and lower reaches of the Yangtze River, China. The gradual habitat loss resulting from wetland degradation may have posed negative... The Hooded Crane (Grus monacha) is a waterbird wintering in the wetlands of the middle and lower reaches of the Yangtze River, China. The gradual habitat loss resulting from wetland degradation may have posed negative effects on the structure of our wintering populations. For its effective protection, it is important to conduct an intensive study on the genetic structure of this population. A total of 221 faecal samples, nine feather samples and four muscle samples of Hooded Cranes from four wintering populations, i.e., from Caizi Lake and Shengjin Lake in Anhui, Poyang Lake in Jiangxi and Chongming Dongtan in Shanghai, were collected for this study. Full-length 1103–1104 bp mtDNA D-loop sequences from 72 samples were amplified using PCR. Based on our amplified D-loop sequences and the sequences of two individual birds obtained from GenBank (AB017625 and AB023813), we analyzed the genetic structure of these four wintering Hooded Crane populations. Twenty six variable sites were found among 72 target sequences in the four wintering populations and 23 haplotypes were defined. Genetic diversity analyses showed that the haplotype diversity of Hooded Cranes was 0.823 ± 0.042 with a nucleotide diversity of 0.00157 ± 0.00021. The FST values of the four populations show that there is no significant genetic differentiation among the populations of Hooded Cranes wintering in the middle and lower reaches of the Yangtze River. Tajima’s D and Fu’s tests suggest that the Hooded Crane populations may have experienced population expansion in their evolutionary history. 展开更多
关键词 grus monacha HAPLOTYPE genetic structure faecal DNA population expansion
原文传递
Status and behavior of the Black-necked Crane (Grus nigricollis) in the Altun Mountain Reserve,Xinjiang
3
作者 张同 马鸣 +2 位作者 丁鹏 徐峰 Paul J.BUZZARD 《Chinese Birds》 2012年第3期199-205,共7页
The Black-necked Crane (Grus nigricollis) is an endemic species of the Qinghai-Tibetan Plateau,with its population in Xinjiang largely found in the Kunlun and Altun mountain regions.A survey of the distribution,popula... The Black-necked Crane (Grus nigricollis) is an endemic species of the Qinghai-Tibetan Plateau,with its population in Xinjiang largely found in the Kunlun and Altun mountain regions.A survey of the distribution,population dynamics,flock size and behavior was conducted in Altun Mountain National Natural Reserve from September to November 2011.We investigated the size and distribution of its population in this area with a sample spot survey and by direct counting.We found Black-necked Cranes on the wetlands of Wuzunxiaoer,Yusup Aleksei,Yaziquan,Qimantag,Tula Ranch and elsewhere,where we recorded 126 individual birds of the species in Yixiekepati (37° 15′-37°23′N,90°11′-90°20′E,elevation 3903 m),the largest population we have observed in this area.In the reserve,the population consists of about 180-200 birds.Combined with previous records,we conclude that more than 260 Black-necked Cranes live in Xinjiang.Cranes gathered conspicuously in the middle of October where the highest number of birds appeared on 29 October.All of the cranes had migrated out of this area by 6 November.Time budgets and diurnal behavior rhythms of Blacknecked Cranes were observed in the Yixiekepati wetland at daylight (from 06:00 to 18:00).Foraging was the most prevalent type of behavior during the autumn period,accounting for 58.9% of the diurnal time budget,followed by preening (13.2%),vigilance (9.5%),walking (8.2%),flying (3.5%),resting (3.4%),chirping (2.7%) and other types of behavior (0.6%). 展开更多
关键词 Black-necked Crane grus nigricollis distribution population size behavior rhythm XINJIANG
原文传递
Effect of water level fluctuations on temporal-spatial patterns of foraging activities by the wintering Hooded Crane(Grus monacha) 被引量:8
4
作者 Dongmei Zhang Lizhi Zhou Yunwei Song 《Chinese Birds》 CSCD 2015年第3期169-177,共9页
Background:The Yangtze River floodplain provides important wintering habitats for Hooded Cranes(Grus monacha) in China.Fluctuations in the water level change foraging habitat and food availability,affecting their temp... Background:The Yangtze River floodplain provides important wintering habitats for Hooded Cranes(Grus monacha) in China.Fluctuations in the water level change foraging habitat and food availability,affecting their temporal-spatial patterns of foraging activities.It is of considerable importance to investigate the effect of these fluctuations on food availability for wintering Hooded Cranes and their foraging response to these changes.Understanding their behavior patterns is beneficial in protecting the wintering crane population and restoring their wintering habitats.Methods:A field survey of the winter behavior of cranes was carried out at Shengjin Lake from November in 2013 to April in 2014.Habitat variables,as well as the spatial distribution and behavior patterns of wintering cranes at their foraging sites during five stages of water level fluctuation were collected.Based on this data we analyzed the relationship of foraging behavior relative to water level fluctuations and habitat types.Results:The foraging habitats used by Hooded Cranes varied at the different water level stages.As the water level decreased,the use of meadows and mudflats increased.When the water dropped to its lowest level,the use by the Hooded Crane in the mudflats reached a peak.There were statistically significant differences in time budget in the three types of habitats over the five stages of the water level.In the mudflats,the foraging behavior and maintenance behavior varied significantly with the water level,while the alert behavior showed little variation.Analysis of a generalized linear model showed that the five water level stages and three habitat types had a significant effect on foraging behavior,while the combined effect of these two variables was significant on the foraging time budget and the length of foraging activity of the Hooded Crane.Conclusions:With the decrease in the water level,the use of mudflats by Hooded Cranes increased correspondingly.Food availability in different habitats was affected by changes in the water level.The Hooded Crane adjusted its foraging patterns and made full use of the three available types of habitat in order to acquire enough food in response to fluctuations in the water level. 展开更多
关键词 Water level FLUCTUATIONS FORAGING activity Temporal-spatial pattern Hooded CRANE grus monacha
原文传递
Effects of variation in food resources on foraging habitat use by wintering Hooded Cranes(Grus monacha) 被引量:8
5
作者 Meng Zheng Lizhi Zhou +1 位作者 Niannian Zhao Wenbin Xu 《Chinese Birds》 CSCD 2015年第3期186-195,共10页
Background:The ideal habitat use of waterbirds can be considered to be fixed,but current habitat use depends on environmental conditions,especially those of food characteristics,considered crucial to their use of habi... Background:The ideal habitat use of waterbirds can be considered to be fixed,but current habitat use depends on environmental conditions,especially those of food characteristics,considered crucial to their use of habitats.Understanding how waterbirds respond to variation in food availability at degraded wetland sites and change their habitat use patterns over spatial and temporal scales should direct future conservation planning.The objectives of this study were to identify these spatial-temporal foraging habitat use patterns of Hooded Cranes(Grus monacha)and their relationship with food characteristics in the severely degraded wetlands of the Shengjin and Caizi lakes along with the Yangtze River floodplain.Methods:We investigated the changes in food characteristics,relative abundance and density of Hooded Cranes in various habitat types across three winter periods from November 2012 to April 2013.We examined the effect of these winter periods and habitat types on the pattern of use by the cranes and explored the relationship between these patterns and food characteristics using linear regression.Results:The food characteristics and habitat use clearly changed over spatial-temporal scales.In the early and mid-winter periods,the most abundant,accessible and frequented food resources were found in paddy fields,while in the late period the more abundant food were available in meadows,which then replaced the paddy fields.There were fewer effects of winter periods,habitat types and their interactions on habitat use patterns except for the effect of habitat types on the relative abundance,determined as a function of food abundance,but independent of food depth and sediment permeability.Conclusions:In response to the degradation and loss of lake wetlands,the cranes shifted their habitat use patterns by making tradeoffs between food abundance and accessibility over spatial-temporal scales that facilitated their survival in the mosaic of these lake wetlands. 展开更多
关键词 Hooded Crane(grus monacha) HABITAT use FOOD characteristics WINTERING ecology FORAGING HABITAT
原文传递
Nest site selection of white-naped crane(Grus vipio) at Zhalong National Nature Reserve, Heilongjiang, China 被引量:3
6
作者 Qing-ming WU Hong-fei ZOU Jian-zhang MA 《Journal of Forestry Research》 SCIE CAS CSCD 2014年第4期947-952,共6页
White-naped crane (Grus vipio) is a globally threatened spe- cies. It is very important to analyze its nest site selection in circum- stances where there are multiple disturbances, and also helpful to accu- mulate v... White-naped crane (Grus vipio) is a globally threatened spe- cies. It is very important to analyze its nest site selection in circum- stances where there are multiple disturbances, and also helpful to accu- mulate valuable information about this threatened species and supply scientific suggestions for conservation and management. We studied nest site selection and the effects of environmental variables on nesting habits of white-naped crane at Zhalong National Nature Reserve, Qiqihar City, Heilongiiang, China, during March-May of 2002-2008. White-naped crane responded and adapted to changes in the quality of the spatial environments of landscape and microhabitat under multiple environ- mental disturbances. Nest site selection included two scales and two choices, namely the choice of nest site habitat type within the macro-habitat scale and nest site micro-habitat selection within the mi- cro-habitat scale. Nest sites were recorded only in reed marshes. The choice of nest site micro-habitat included three basic elements and six factors, namely incubation element (nest parameters factor, incubation temperature factor and incubation humidity factor), safety element (pro- tection factor and concealment factor), and food element (water factor). Water, remnant reed clusters, and fire were major resource management challenges during the breeding period for the white-naped crane in this Reserve. 展开更多
关键词 factor analysis environmental variable grus vipio nestsite selection pattern Zhalong Nature Reserve
在线阅读 下载PDF
Monitoring population dynamics of the migratory Red-crowned Crane (Grus japonensis) at Shuangtaihekou National Nature Reserve,northeastern China,from 1991 to 2012 被引量:1
7
作者 李玉祥 宋常站 +4 位作者 杨玉成 李晓静 李风丽 黄杰 黄秀艳 《Chinese Birds》 2012年第3期225-229,共5页
We studied migratory Red-crowned cranes from 1991-2012 in order to track population dynamics and trends for these cranes over time at Shuangtaihekou,northeastern China.Our results show that the earliest arrival date o... We studied migratory Red-crowned cranes from 1991-2012 in order to track population dynamics and trends for these cranes over time at Shuangtaihekou,northeastern China.Our results show that the earliest arrival date of the Red-crowned Crane (Grus japonensis) was usually around 1 March in the spring and the latest departure date of the cranes around 8 November.The peak period for the number of cranes observed in the spring was 10-20 March over these 22 years.The staging period in the spring is about 25 days.Comparing the highest number of Red-crowned cranes observed in the spring,the migratory population was around 400 birds in the 1990s,then,increased to over 800 birds in 2000.For a short while,the population maintained itself around 600-800 birds.After that,the population started to decline,until during the most recent years,we observed around 400 birds at peak times of each season.Our results match a similar population change pattern observed in the wintering area for a western population,which has been declining since 2000.There is an urgent need to reinforce protection and management,as well as to maintain and improve the quality of their habitats. 展开更多
关键词 grus japonensis MIGRATION population dynamics Shuangtaihekou STOPOVER
原文传递
Long-term monitoring data reveal effects of age,population density,and environmental aspects on hatching success of Common Cranes(Grus grus)
8
作者 Isabel Barwisch Wolfgang Mewes Angela Schmitz Ornes 《Avian Research》 SCIE CSCD 2022年第3期347-355,共9页
Facing climate and land use change,a species’ability to successfully adapt to changing environments is crucial for its survival.Extensive drainage and intensification of agriculture and forestry set wetlands and asso... Facing climate and land use change,a species’ability to successfully adapt to changing environments is crucial for its survival.Extensive drainage and intensification of agriculture and forestry set wetlands and associated species at risk of population declines.The population of Common Cranes(Grus grus)has experienced considerable fluctuations over the last century.Despite increasing population numbers,hatching success seemed to have decreased over the last years.The aim of this study was to identify factors influencing hatching success and nest survival of Common Cranes based on analyses of long-term individual-based monitoring data from northeastern Germany and evaluate the species ability to adapt to changing environments.Hatching success decreased over the course of the study period from 0.75 to 0.55.Surprisingly,nest survival and hatching success did not vary across different nesting habitats,whereas factors such as female age,timing of nest initiation and breeding pair density were found to have significant effects on hatching success.Older females showed higher hatching success,even though the proportion of unhatched eggs was highest in females aged 20 years or older.Early nest initiation had a positive effect on hatching success.Water levels are more favorable early in the nesting season,whereas increasing evaporation with time causes water levels to decrease,granting easier access for predators.Independently of female age,hatching success decreased with increasing numbers of breeding pairs within a 2-km radius around a nesting site.High population densities intensify competition for resources and promote intraspecific interactions,affecting reproductive outcome negatively.This study gives first insights into mechanisms behind population regulation in Common Cranes,highlighting the importance of population dynamics and individual features.We suggest to further investigate density dependent effects including landscape and habitat features as well as reproductive success in terms of chick survival,since successfully raising juveniles is crucial for a species survival. 展开更多
关键词 ADAPTATION Age CRANES grus grus HABITAT Nest survival Population density REPRODUCTION
在线阅读 下载PDF
MICROSTRUCTURE OF FEATHER OF RED CROWNED CRANE (GRUS JAPONENSIS)
9
作者 景松岩 杭馥兰 +2 位作者 李悦 刘家琪 王学君 《Journal of Northeast Forestry University》 SCIE CAS CSCD 1995年第2期71-76,共6页
Through scanning electron microscope, various trpes of feathers in various parts of red-crown crane--a kind of rare and precious bird were observed micrographically. It was shown in the result that in different type o... Through scanning electron microscope, various trpes of feathers in various parts of red-crown crane--a kind of rare and precious bird were observed micrographically. It was shown in the result that in different type of feather, the barbules show a great variation in their morphology and structure. This observation can be a basic reference for the taxonomy using the feather structure and the study of systematic relationship. 展开更多
关键词 grus japonensis MICROSTRUCTURE FEATHER
在线阅读 下载PDF
多关系和时间增强的知识追踪模型 被引量:2
10
作者 张维 罗佩华 +2 位作者 龚中伟 李志新 宋玲玲 《计算机应用研究》 北大核心 2025年第3期728-734,共7页
现有知识追踪方法未能深入探索知识点间多种关系并同时考虑知识相互作用和时间对知识状态的影响。鉴于此,从知识间多种关系和学习遗忘规律两方面改进知识追踪模型,提出多关系和时间增强的知识追踪模型(MRTKT)。首先,根据认知同化理论丰... 现有知识追踪方法未能深入探索知识点间多种关系并同时考虑知识相互作用和时间对知识状态的影响。鉴于此,从知识间多种关系和学习遗忘规律两方面改进知识追踪模型,提出多关系和时间增强的知识追踪模型(MRTKT)。首先,根据认知同化理论丰富了知识间关系,使用统计学方法构建了包含上位学习、下位学习以及并列组合学习三种关系的知识结构图;其次,对知识间相互作用进行建模,根据上述三种关系聚集节点特征,使得模型可以更好地模拟知识间的影响传播行为;然后,构建融入三种时间信息的GRU门更新学生知识状态,以模拟学习和遗忘对知识状态的影响,使得各节点特征包含知识间相互作用信息和时间信息,为预测学习者答题表现提供更全面丰富的信息。在多个公开数据集上进行实验,结果表明MRTKT比现有模型具有更优越的性能以及更好的可解释性。 展开更多
关键词 知识追踪 门控图神经网络 GRU 知识间多种关系 学习遗忘
在线阅读 下载PDF
认知结构动态建模和遗忘显式计算的知识追踪模型
11
作者 张维 罗佩华 +2 位作者 李志新 龚中伟 宋玲玲 《计算机科学与探索》 北大核心 2025年第10期2683-2696,共14页
知识追踪任务旨在根据学生历史学习记录来估计学生的知识掌握程度,其中准确建模学习和遗忘行为是精准刻画学生知识状态的关键。现有方法通常基于学生静态知识结构对学习行为进行建模,并通过拼接时间特征和交互信息的隐式方式以改进遗忘... 知识追踪任务旨在根据学生历史学习记录来估计学生的知识掌握程度,其中准确建模学习和遗忘行为是精准刻画学生知识状态的关键。现有方法通常基于学生静态知识结构对学习行为进行建模,并通过拼接时间特征和交互信息的隐式方式以改进遗忘建模。然而,学生的认知结构是随时间变化的,且隐式处理时间特征不能充分利用时间信息。为了解决上述问题,提出了一种认知结构动态建模和遗忘显式计算的知识追踪方法(CSFKT),以更好地捕获学生知识状态变化情况。该方法根据学生答题反馈,使用门控循环单元(GRU)对学生认知结构的邻接矩阵进行更新,构建动态变化的学生认知结构图;基于该认知结构图,利用图神经网络的邻域聚合策略建模知识点相互作用过程;提出一种遗忘显式计算方法,利用间隔时间和遗忘曲线公式显示计算记忆保留概率及衰减后的知识状态,再使用GRU获得当前时刻的知识状态,并预测学生正确作答的概率。在三个真实数据集上进行了大量实验,结果表明CSFKT不仅可以建模动态的认知结构也可以显式建模学生遗忘行为,而且具有优越的性能以及良好的可解释性。 展开更多
关键词 知识追踪 图神经网络(GNN) 门控循环单元(GRU) 认知结构 遗忘曲线
在线阅读 下载PDF
基于GASF多通道图像时序融合的高速列车横向减振器故障诊断
12
作者 李刚 秦永峰 齐金平 《振动与冲击》 北大核心 2025年第15期144-152,191,共10页
由于高速列车在运行的过程中悬挂系统产生的振动信号是典型的复杂度高,耦合性和不确定性强的非线性信号,为弥补一维信号在故障诊断时的局限性,利用格拉姆角场(Gramian angular field,GAF)处理时间序列信号的敏感性以及对非线性信号的适... 由于高速列车在运行的过程中悬挂系统产生的振动信号是典型的复杂度高,耦合性和不确定性强的非线性信号,为弥补一维信号在故障诊断时的局限性,利用格拉姆角场(Gramian angular field,GAF)处理时间序列信号的敏感性以及对非线性信号的适应性,提出了一种基于一维(1D)时序信号和二维(2D)格拉姆角和场(Gramian angular summation field,GASF)特征融合的卷积神经网络结合门控循环单元网络融合多头自注意力机制(1D-2D-CNN-GRU-MSA)的故障诊断方法。首先,将一维的时序信号编码为二维的GASF图,再分别将一维的时序信号与二维的GASF图同时送入到两条并行的网络支路中,其中:一路为图像输入经卷积神经网络(convolutional neural networks,CNN)提取GASF图像的特征;另一路将一维的故障波形直接输入经门控循环网络单元(gated recurrent unit,GRU)提取时序特征,通过多头自注意力机制(multi-head self-attention,MSA)将二维图像特征和一维时序特征进行特征重点强化并降维融合,最后通过Softmax层对高速列车横向减振器故障进行分类。仿真试验证明,不同工况下1D-2D-CNN-GRU-MSA模型比两种主流模型进行高速列车横向减振器故障识别的准确率高。 展开更多
关键词 高速列车 格拉姆角和场(GASF) 卷积神经网络(CNN) 多头自注意力机制(MSA) 门控循环单元(GRU)
在线阅读 下载PDF
基于多尺度动态卷积和GRU的轴承故障诊断
13
作者 董绍江 彭银山 +1 位作者 邹松 黄翔 《组合机床与自动化加工技术》 北大核心 2025年第3期150-154,共5页
针对传统轴承故障诊断过程中忽略轴承振动信号的关联时间维度信息的问题,提出了基于多尺度动态扩张卷积神经网络(MSDDCNN)和门控循环单元网络(GRU)的轴承故障诊断方法。首先利用不同尺寸宽卷积核从各个维度捕捉振动信号多维特征以增大... 针对传统轴承故障诊断过程中忽略轴承振动信号的关联时间维度信息的问题,提出了基于多尺度动态扩张卷积神经网络(MSDDCNN)和门控循环单元网络(GRU)的轴承故障诊断方法。首先利用不同尺寸宽卷积核从各个维度捕捉振动信号多维特征以增大感受野;其次引入动态加权层自适应选择卷积核尺度的大小并自动地给予特征序列中的不同部分不同的权重,更加充分提升特征表示的能力;最后利用门控循环单元充分提取振动信号中不同尺度的时序特征,以加强各个维度间前后时间维度关联信息。实验结果表明,所提方法在PU和JNU公开数据集上平均准确率分别为98.79%和98.65%。为验证所提网络模型诊断有效性,所提方法在某公司自制的轴承故障数据集(CME)也表现出较高的准确率和较大抗噪声能力,为有效诊断旋转部件故障提供了实际依据。 展开更多
关键词 多尺度 动态卷积 扩张卷积 GRU 故障诊断
在线阅读 下载PDF
人工智能大数据技术在治理虚拟货币刑事犯罪中的应用研究
14
作者 刘欢 肖蔚 《现代电子技术》 北大核心 2025年第16期167-171,共5页
针对虚拟货币刑事犯罪治理需求,提出一种基于动态图卷积网络的增强识别模型(RGCN)。创新性地在图嵌入层设计中融合原始特征与跨时特征的拼接机制,有效缓解了卷积冗余导致的信息流失问题;采用随机森林集成学习策略优化图嵌入层特征表达,... 针对虚拟货币刑事犯罪治理需求,提出一种基于动态图卷积网络的增强识别模型(RGCN)。创新性地在图嵌入层设计中融合原始特征与跨时特征的拼接机制,有效缓解了卷积冗余导致的信息流失问题;采用随机森林集成学习策略优化图嵌入层特征表达,显著提升模型识别精度。所构建的3层RGCN框架通过时空特征融合策略降低计算开销,并重点优化召回率。实验结果表明:LSTM动态图卷积网络在召回率指标上表现最优,为67.6%,较GRU变体提升4.4%;而GRU结构则更侧重精确率优化,精确率达到87.3%。最后,结合当前反洗钱场景对高风险个体监控的迫切需求,验证了所提模型通过强化召回率性能,可有效支撑虚拟货币交易监管实践。 展开更多
关键词 人工智能 虚拟货币 刑事犯罪 动态图卷积网络 LSTM GRU 召回率优化
在线阅读 下载PDF
利用GRU双分支信息协同增强的长尾推荐模型
15
作者 钱忠胜 肖双龙 +2 位作者 朱辉 王晓闻 刘金平 《计算机科学与探索》 北大核心 2025年第2期476-489,共14页
长尾现象在序列推荐系统中长期存在,包括长尾用户和长尾项目两个方面。虽然现有许多研究缓解了序列推荐系统中的长尾问题,但大部分只是单方面地关注长尾用户或长尾项目。然而,长尾用户和长尾项目问题常常同时存在,只考虑其中一方会导致... 长尾现象在序列推荐系统中长期存在,包括长尾用户和长尾项目两个方面。虽然现有许多研究缓解了序列推荐系统中的长尾问题,但大部分只是单方面地关注长尾用户或长尾项目。然而,长尾用户和长尾项目问题常常同时存在,只考虑其中一方会导致另一方性能不佳,且未关注到长尾用户、长尾项目各自的信息匮乏问题。提出一种利用GRU双分支信息协同增强的长尾推荐模型(long-tail recommendation model utilizing gated recurrent unit dualbranch information collaboration enhancement,LT-GRU),从用户与项目两个方面共同缓解长尾问题,并通过协同增强的方式丰富长尾信息。该模型由长尾用户和长尾项目双分支组成,每个分支分别负责各自的信息处理,并相互训练以充实另一方的信息。同时,引入一种偏好机制,通过演算用户与项目的影响因子,以动态调整用户偏好与项目热度,进一步缓解长尾推荐中信息不足问题。在Amazon系列的6个真实数据集上与6种经典模型进行实验对比,相较于长尾推荐模型中最优的结果,所提模型LT-GRU在HR与NDCG两个指标上分别平均提高2.49%、3.80%。这表明,在不牺牲头部用户和热门项目推荐性能的情况下,有效地缓解了长尾用户和长尾项目问题。 展开更多
关键词 推荐系统 长尾推荐 信息协同增强 门控循环单元(GRU)
在线阅读 下载PDF
一种真实人体无触重建的并行GRU网络算法
16
作者 刘岩 尹伟石 《黑龙江大学自然科学学报》 2025年第2期154-161,共8页
通过真实特征参数重建人体是虚拟现实领域的重要问题。为了实现高精度人体重建并降低计算复杂度,构建并行门控循环单元(Gated recurrent unit,GRU)网络建立人体特征参数和远场数据之间的关系,使用Fourier方法通过远场数据重建人体形状... 通过真实特征参数重建人体是虚拟现实领域的重要问题。为了实现高精度人体重建并降低计算复杂度,构建并行门控循环单元(Gated recurrent unit,GRU)网络建立人体特征参数和远场数据之间的关系,使用Fourier方法通过远场数据重建人体形状。从低维度数据向高维度数据映射会导致模型泛化能力减弱,通过并行GRU网络实现高维远场数据快速计算,有效建立低维人体特征参数和高维远场数据之间的关系。实验结果表明,该方法可以实现三维人体重建,在真实数据的数据集上具有较低的平均误差。 展开更多
关键词 三维人体重建 并行GRU网络 远场数据 人体特征参数
在线阅读 下载PDF
沉管隧道长期沉降预测模型研究
17
作者 胡晓勇 郑万坤 《建筑机械》 2025年第3期211-215,220,共6页
文章针对沉管隧道沉降监测数据非线性、非平稳的特点,提出1种基于STL分解的GRU-ARIMA组合预测模型,可更好地兼顾沉管隧道沉降长期发展趋势和波动特征。利用STL分解将沉降监测数据分解为趋势分量、周期分量和噪音分量,采用GRU循环神经网... 文章针对沉管隧道沉降监测数据非线性、非平稳的特点,提出1种基于STL分解的GRU-ARIMA组合预测模型,可更好地兼顾沉管隧道沉降长期发展趋势和波动特征。利用STL分解将沉降监测数据分解为趋势分量、周期分量和噪音分量,采用GRU循环神经网络对趋势分量和噪音分量进行预测,周期分量采用ARIMA模型预测,将预测结果相加获得总的预测值。以车陂路-新滘东路隧道工程为例,选择代表性监测点J1、J2的沉降监测数据,划分训练集和测试集,通过对比测试集实测值和组合模型预测值以验证模型的合理性,同时与单一GRU循环神经网络预测值进行对比以验证组合模型的优势。结果表明:组合模型测试集预测结果与实测值相比较为吻合,能够用于实际工程中,同时相对于单一的GRU模型具有明显的优势。 展开更多
关键词 沉管隧道 沉降监测 STL分解 GRU循环神经网络 ARIMA模型
在线阅读 下载PDF
基于GRU神经网络的电商网络恶意流量检测研究
18
作者 王自豪 谢丽 李俊 《微型电脑应用》 2025年第7期186-189,共4页
电商网络的恶意流量中存在长序列、多任务等特征,导致检测精度下降,对此,研究基于门控循环单元(GRU)神经网络的电商网络恶意流量检测方法。在相空间重构过程中,引入小波变换消除电商网络流量中存在的噪声。采用主成分分析法对去噪后的... 电商网络的恶意流量中存在长序列、多任务等特征,导致检测精度下降,对此,研究基于门控循环单元(GRU)神经网络的电商网络恶意流量检测方法。在相空间重构过程中,引入小波变换消除电商网络流量中存在的噪声。采用主成分分析法对去噪后的电商网络流量进行特征提取。将经过特征提取的电商网络流量输入GRU神经网络,通过训练和优化神经网络得到电商网络恶意流量检测结果。实验结果表明,所提方法的电商网络流量处理效果好,恶意流量类别检测精度高达到99%,最短检测时间仅为234 s,具有一定的技术水平与实用性。 展开更多
关键词 GRU神经网络 相空间重构 小波变换 主成分分析 网络恶意流量检测
在线阅读 下载PDF
基于改进DQN的Kinova三指机械臂抓取策略优化
19
作者 韩玉兰 李聪 +1 位作者 柳子涵 田旭航 《物联网技术》 2025年第23期52-58,共7页
针对深度Q网络(DQN)在复杂控制任务实施过程中面临的Q值估计过高、计算复杂、记忆功能缺失和探索不足等问题,提出一种改进的DQN模型,旨在提升Kinova三指机械臂的抓取性能。在PyBullet仿真环境中获取动态图像输入,并引入门控循环单元(GRU... 针对深度Q网络(DQN)在复杂控制任务实施过程中面临的Q值估计过高、计算复杂、记忆功能缺失和探索不足等问题,提出一种改进的DQN模型,旨在提升Kinova三指机械臂的抓取性能。在PyBullet仿真环境中获取动态图像输入,并引入门控循环单元(GRU)以捕捉和处理时间序列信息,从而增强模型对历史状态的记忆能力。同时,采用NoisyLinear层替代传统epsilon-greedy策略中的随机噪声,显著提升了策略的探索效率与多样性。在模型训练过程中,结合噪声线性探索机制和经验回放机制,实验结果表明,改进后的模型可显著提高机械臂的抓取成功率。 展开更多
关键词 DQN GRU Q值 NoisyLinear 机械臂抓取 经验回放
在线阅读 下载PDF
融合GRU和改进蚁群算法的火灾逃生路径规划方法
20
作者 赵峰 姜孝文 《信息技术与信息化》 2025年第1期65-69,共5页
高温环境下,传感器节点可能会出现工作状态异常的情况,由此引发误报、漏报等一系列问题,文章提出了融合门控循环单元神经网络(gated recurrent unit,GRU)和改进蚁群算法的室内建筑火灾逃生路径规划研究方法。根据火灾态势信息,建立GRU... 高温环境下,传感器节点可能会出现工作状态异常的情况,由此引发误报、漏报等一系列问题,文章提出了融合门控循环单元神经网络(gated recurrent unit,GRU)和改进蚁群算法的室内建筑火灾逃生路径规划研究方法。根据火灾态势信息,建立GRU时序预测模型,实现对节点异常情况下火场内温度、CO浓度等态势信息的预测,并基于预测结果,利用改进的蚁群算法动态规划逃生路径,得到传感器工作异常情况下的火灾最佳安全逃生路径。研究结果表明,该方法在不同火灾时期都能有效规划出最佳疏散路径,具有实际应用价值。 展开更多
关键词 GRU神经网络模型 火灾疏散路径动态规划 蚁群算法
在线阅读 下载PDF
上一页 1 2 39 下一页 到第
使用帮助 返回顶部