期刊文献+
共找到1,360篇文章
< 1 2 68 >
每页显示 20 50 100
Fault detection observer design for networked control system with long time-delays and data packet dropout 被引量:3
1
作者 Xuan Li Xiaobei Wu +1 位作者 Zhiliang Xu Cheng Huang 《Journal of Systems Engineering and Electronics》 SCIE EI CSCD 2010年第5期877-882,共6页
Focusing on the networked control system with long time-delays and data packet dropout,the problem of observerbased fault detection of the system is studied.According to conditions of data arrival of the controller,th... Focusing on the networked control system with long time-delays and data packet dropout,the problem of observerbased fault detection of the system is studied.According to conditions of data arrival of the controller,the state observers of the system are designed to detect faults when they occur in the system.When the system is normal,the observers system is modeled as an uncertain switched system.Based on the model,stability condition of the whole system is given.When conditions are satisfied,the system is asymptotically stable.When a fault occurs,the observers residual can change rapidly to detect the fault.A numerical example shows the effectiveness of the proposed method. 展开更多
关键词 networked control system fault detection long timedelays data packet dropout.
在线阅读 下载PDF
Oil-gas reservoir in the Mesozoic strata in the Chaoshan depression,northern South China Sea:a new insight from long off set seismic data 被引量:1
2
作者 Tao XING Guangjian ZHONG +2 位作者 Wenhuan ZHAN Zhongquan ZHAO Xi CHEN 《Journal of Oceanology and Limnology》 SCIE CAS CSCD 2022年第4期1377-1387,共11页
The Chaoshan depression,a Mesozoic basin in the Dongsha sea area,northern South China Sea,is characterized by well-preserved Mesozoic strata,being good conditions for oil-gas preservation,promising good prospects for ... The Chaoshan depression,a Mesozoic basin in the Dongsha sea area,northern South China Sea,is characterized by well-preserved Mesozoic strata,being good conditions for oil-gas preservation,promising good prospects for oil-gas exploration.However,breakthrough in oil-gas exploration in the Mesozoic strata has not been achieved due to less seismic surveys.New long-off set seismic data were processed that acquired with dense grid with single source and single cable.In addition,the data were processed with 3D imaging method and fi ner processing was performed to highlight the target strata.Combining the new imaging result and other geological information,we conducted integrated interpretation and proposed an exploratory well A-1-1 for potential hydrocarbon.The result provides a reliable basis for achieving breakthroughs in oil and gas exploration in the Mesozoic strata in the northern South China Sea. 展开更多
关键词 Chaoshan depression Mesozoic strata oil and gas exploration long off set seismic data integrated interpretation exploratory well
在线阅读 下载PDF
LncPipe" A Nextflow-based pipeline for identification and analysis of long non-coding RNAs from RNA-Seq data 被引量:2
3
作者 Qi Zhao Yu Sun +4 位作者 Dawei Wang Hongwan Zhang Kai Yu Jian Zheng Zhixiang Zuo 《Journal of Genetics and Genomics》 SCIE CAS CSCD 2018年第7期399-401,共3页
Long noncoding RNAs (IncRNAs) have been increasingly implicated in a variety of human diseases, including autoimmune disease (Wu et al., 2015), neurodegenerative diseases (Wapinski and Chang, 2011) and cancer (... Long noncoding RNAs (IncRNAs) have been increasingly implicated in a variety of human diseases, including autoimmune disease (Wu et al., 2015), neurodegenerative diseases (Wapinski and Chang, 2011) and cancer (Huarte, 2015). Due to recent advances in next-generation sequencing technologies, tens of thousands of lncRNAs have been identified and annotated, a number of them have been proven to be functional in diverse biological processes through various mechanisms. 展开更多
关键词 LncPipe" A Nextflow-based pipeline IDENTIFICATION analysis of long non-coding RNAs RNA-Seq data
原文传递
Long-range correlation analysis of urban traffic data
4
作者 盛鹏 王俊峰 +1 位作者 唐铁桥 赵树龙 《Chinese Physics B》 SCIE EI CAS CSCD 2010年第8期55-64,共10页
This paper investigates urban traffic data by analysing the long-range correlation with detrended fluctuation analysis. Through a large number of real data collected by the travel time detection system in Beijing, the... This paper investigates urban traffic data by analysing the long-range correlation with detrended fluctuation analysis. Through a large number of real data collected by the travel time detection system in Beijing, the variation of flow in different time periods and intersections is studied. According to the long-range correlation in different time scales, it mainly discuss the effect of intersection location in road net, people activity customs and special traffic controls on urban traffic flow. As demonstrated by obtained results, the urban traffic flow represents three-phase characters similar to highway traffic. Moreover, compared by the two groups of data obtained before and after the special traffic restrictions (vehicles with special numbered plates only run in a special workday) enforcement, it indicates that the rules not only reduce the flow but also avoid irregular fluctuation. 展开更多
关键词 urban traffic data long-range correlation detrended fluctuation analysis special traffic restriction
原文传递
Design of Low-Power Data Logger of Deep Sea for Long-Term Field Observation 被引量:1
5
作者 赵伟 陈鹰 +2 位作者 杨灿军 曹建伟 顾临怡 《China Ocean Engineering》 SCIE EI 2009年第1期133-144,共12页
This paper describes the implementation of a data logger for the real-time in-situ monitoring of hydrothermal systems. A compact mechanical structure ensures the security and reliability of data logger when used under... This paper describes the implementation of a data logger for the real-time in-situ monitoring of hydrothermal systems. A compact mechanical structure ensures the security and reliability of data logger when used under deep sea. The data logger is a battery powered instrument, which can connect chemical sensors (pH electrode, H2S electrode, H2 electrode) and temperature sensors. In order to achieve major energy savings, dynamic power management is implemented in hardware design and software design. The working current of the data logger in idle mode and active mode is 15 μA and 1.44 mA respectively, which greatly extends the working time of battery. The data logger has been successftdly tested in the first Sino-American Cooperative Deep Submergence Project from August 13 to September 3, 2005. 展开更多
关键词 data logger low-power design deep sea long-term monitoring
在线阅读 下载PDF
基于RAG的航空维修Long-Term数据提取技术研究
6
作者 赵华蕾 徐豪 +2 位作者 张国兴 陈巍 陈涤非 《长江信息通信》 2025年第8期5-8,共4页
针对航空领域的维修手册和工作记录多存在于Long-term(大表格或长文本)数据中,提出了一种基于RAG的航空维修Long-term数据提取技术方案,该方案将long-term变成short-term再进行检索,能够实现对复杂维修数据的高效管理和优化处理,保持数... 针对航空领域的维修手册和工作记录多存在于Long-term(大表格或长文本)数据中,提出了一种基于RAG的航空维修Long-term数据提取技术方案,该方案将long-term变成short-term再进行检索,能够实现对复杂维修数据的高效管理和优化处理,保持数据信息的完整性,提升航空维修的效率和准确性。 展开更多
关键词 RAG 航空维修 long-TERM SHORT-TERM 数据提取
在线阅读 下载PDF
基于长尾词分布的藏汉机器翻译数据增强方法
7
作者 格桑加措 尼玛扎西 +5 位作者 群诺 嘎玛扎西 道吉扎西 罗桑益西 拉毛吉 钱木吉 《计算机科学》 北大核心 2026年第1期224-230,共7页
现有藏汉机器翻译语料中存在领域数据分布不平衡的问题,导致训练出来的模型对各个领域数据的翻译能力表现不均衡。反向翻译作为一种常见的数据增强方法,通过提供更多样化的伪数据来提高模型的性能。然而,传统的反向翻译方法难以充分考... 现有藏汉机器翻译语料中存在领域数据分布不平衡的问题,导致训练出来的模型对各个领域数据的翻译能力表现不均衡。反向翻译作为一种常见的数据增强方法,通过提供更多样化的伪数据来提高模型的性能。然而,传统的反向翻译方法难以充分考虑数据的领域分布不平衡问题,导致模型在整体性能提升过程中难以提升资源稀缺领域的翻译性能。对此,通过深入分析语料中的长尾词的分布,有针对性地利用现有藏汉双语语料的长尾词来选取单语数据,通过反向翻译构造伪数据进行数据增强操作。这一策略旨在提升藏汉机器翻译模型整体性能的同时,改善数据匮乏领域的翻译性能。实验结果表明,通过充分考虑领域数据不平衡情况,结合长尾词数据增强,能够有效提升机器翻译模型在稀缺领域的翻译性能,为解决领域数据不平衡问题提供了一种有针对性的策略。 展开更多
关键词 长尾词 数据增强 藏汉机器翻译 领域数据不平衡
在线阅读 下载PDF
融合通道注意力与SDFFT的同质多变量时序预测模型
8
作者 王凯欣 杨磊 +3 位作者 赵品杰 江宇函 李康顺 刘启健 《软件导刊》 2026年第1期83-94,共12页
在同质多变量时间序列长期预测任务中,现有模型大多存在着特征提取时未充分利用变量间内在关联信息、对数据的时效性考虑不足、季节性和随机波动信息提取不到位等问题。为此,提出了一种融合多阶段跨时间依赖和拐点信息的同质多变量时间... 在同质多变量时间序列长期预测任务中,现有模型大多存在着特征提取时未充分利用变量间内在关联信息、对数据的时效性考虑不足、季节性和随机波动信息提取不到位等问题。为此,提出了一种融合多阶段跨时间依赖和拐点信息的同质多变量时间序列预测模型(HMTS-TI),设计了RGCA模块,在变量间通道注意力中引入分组排序的策略,将学习到的关联信息转化为注意权重,增强对多变量时间序列变量间相关信息的提取;构建了SDFFT模块,通过子序列分解与快速傅里叶变换,挖掘长期趋势、短期季节模式及随机波动信息,加强跨时间维度依赖特征提取;将拐点数量统计值作为预测调整因子,提取不同时间序列变量中的波动信息,对模型拟合产生正向调节作用。在Traffic、Solar、PeMS04 3个公共同质多变量时间序列数据集上的实验结果表明,相较于对比实验中的最优模型,HMTS-TI在96、192、336和720共4个预测长度上整体均方误差、平均绝对误差分别下降了12.1%和6.8%,决定系数提升了4.1%,具有较高的预测准确率。此外,在具体应用案例中,该模型对广东省内92个观测点的臭氧数据预测长度为336个时间单位任务上的均方误差相比最优对比模型下降了19.0%。 展开更多
关键词 同质多变量时间序列 长期预测 通道注意力 子序列划分 数据分解 快速傅里叶变换
在线阅读 下载PDF
基于散索鞍转动位移的悬索桥主缆力长期监测
9
作者 董皓 单德山 +1 位作者 于伟栋 罗凌峰 《工程科学与技术》 北大核心 2026年第1期233-244,共12页
为解决已有测试设备和方法无法准确、有效测量主缆力的问题,对悬索桥主缆张力长期可靠监测方法展开研究。将悬索桥各结构部件简化为弹簧,组成一个串并联系统等效模型,将缆索刚度视为弹性拉伸刚度和线形变化刚度的串联组合,分别获得抛物... 为解决已有测试设备和方法无法准确、有效测量主缆力的问题,对悬索桥主缆张力长期可靠监测方法展开研究。将悬索桥各结构部件简化为弹簧,组成一个串并联系统等效模型,将缆索刚度视为弹性拉伸刚度和线形变化刚度的串联组合,分别获得抛物线索和悬链线索的纵向刚度,利用能量法并考虑2阶效应的影响,推导桥塔对主缆的约束刚度。在此基础上,根据散索鞍处的平衡原理,结合悬索桥刚度等效简化模型,提出一种利用散索鞍位移确定主缆张力的方法。以某主跨660 m的悬索桥为例,通过有限元建模验证所提方法的准确性和有效性,并进一步用模型对该桥181 d的监测数据进行处理分析。研究表明:散索鞍位移和温度具有明显的线性相关性,温度归一化后的散索鞍位移数据服从正态分布;监测周期内主缆力波动2847 kN,仅为主缆力均值的2.5%;主缆结构健康状况良好,具有足够的承载力,建立的ARIMA(autoregressive integrated moving average)模型能够准确预测主缆力的大小和变化趋势,可为结构的异常和性能退化提供可靠的判断依据。 展开更多
关键词 桥梁工程 大跨悬索桥 主缆张力监测 数据处理分析 时间序列模型
在线阅读 下载PDF
The Uses and Benefits of Big Data for Geological Surveys 被引量:2
10
作者 Michael H.STEPHENSON 《Acta Geologica Sinica(English Edition)》 SCIE CAS CSCD 2019年第S01期64-65,共2页
’Long tail’data is the difficult-to-get-at data that sits in libraries,institutes and on the computers of individual scientists.Informatics specialists like to contrast it with the smaller number of large,more acces... ’Long tail’data is the difficult-to-get-at data that sits in libraries,institutes and on the computers of individual scientists.Informatics specialists like to contrast it with the smaller number of large,more accessible data sets(e.g.Sinha et al.,2013).The name’long tail’derives from graphs drawn of the size of data sets against their number:there are relatively few large datasets and a lot of smaller ones. 展开更多
关键词 long TAIL data informatics BIOSTRATIGRAPHY GEOLOGICAL survey
在线阅读 下载PDF
Long-term survival of early-stage non-small cell lung cancer patients who underwent robotic procedure:a propensity score-matched study 被引量:4
11
作者 Hao-Xian Yang 《Chinese Journal of Cancer》 SCIE CAS CSCD 2016年第7期339-341,共3页
Background:In the past decade,many researchers focused on to robot-assisted surgery.However,on long-term outcomes for patients with early-stage non-small cell lung cancer(NSCLC),whether the robotic procedure is superi... Background:In the past decade,many researchers focused on to robot-assisted surgery.However,on long-term outcomes for patients with early-stage non-small cell lung cancer(NSCLC),whether the robotic procedure is superior to video-assisted thoracic surgery(VATS) and thoracotomy is unclear.Nonetheless,in the article titled "Long-term survival based on the surgical approach to lobectomy for clinical stage I non-small cell lung cancer:comparison of robotic,video assisted thoracic surgery,and thoracotomy lobectomy" by Yang et al.that was recently published in Annals of Surgery,the authors provided convincing evidence that the robotic procedure results in similar long-term survival as compared with VATS and thoracotomy.Minimally invasive procedures typically result in shorter lengths of hospital stay,and the robotic procedure in particular results in superior lymph node assessment.Main body:Our propensity score-matched study generated high-quality data.Based on our findings,we see promise in expanding patient access to robotic lung resections.In this study,propensity score matching minimized the bias involved between groups.Nevertheless,due to its retrospective nature,bias may still exist.Currently,the concept of rapid rehabilitation is widely accepted,and it is very difficult to set up a randomized controlled trial to compare robotic,VATS,and thoracotomy procedures for the treatment of NSCLC.Therefore,to overcome this limitation and to minimize bias,the best approach is to use a registry and prospectively collected,propensity score-matched data.Conclusions:Robotic lung resections result in similar long-term survival as compared with VATS and thoracotomy.Robot-assisted and VATS procedures are associated with short lengths of hospital stay,and the robotic procedure in particular results in superior lymph node assessment.Considering the alarming increase in the incidence of lung cancer in China,a nationwide database of prospectively collected data available for clinical research would be especially important. 展开更多
关键词 非小细胞肺癌 机器人手术 生存率 患者 配研 评分 早期 倾向性
暂未订购
Long-Term Trends in Photosynthetically Active Radiation in Beijing 被引量:1
12
作者 胡波 王跃思 刘广仁 《Advances in Atmospheric Sciences》 SCIE CAS CSCD 2010年第6期1380-1388,共9页
A long-term dataset of photosynthetically active radiation (Qp) is reconstructed from a broadband global solar radiation (Rs) dataset through an all-weather reconstruction model. This method is based on four years... A long-term dataset of photosynthetically active radiation (Qp) is reconstructed from a broadband global solar radiation (Rs) dataset through an all-weather reconstruction model. This method is based on four years' worth of data collected in Beijing. Observation data of Rs and Qp from 2005-2008 are used to investigate the temporal variability of Qp and its dependence on the clearness index and solar zenith angle. A simple and effcient all-weather empirically derived reconstruction model is proposed to reconstruct Qp from Rs. This reconstruction method is found to estimate instantaneous Qp with high accuracy. The annual mean of the daily values of Qp during the period 1958-2005 period is 25.06 mol m-2 d-1. The magnitude of the long-term trend for the annual averaged Qp is presented (-0.19 mol m-2 yr-1 from 1958-1997 and -0.12 mol m-2 yr-1 from 1958-2005). The trend in Qp exhibits sharp decreases in the spring and summer and more gentle decreases in the autumn and winter. 展开更多
关键词 photosynthetically active radiation historical data reconstruction long-term trends
在线阅读 下载PDF
Analysis of long non-coding RNA expression profiles in gastric cancer 被引量:33
13
作者 Wei-Jun Cao Hai-Lu Wu +2 位作者 Bang-Shun He Yu-Shu Zhang Zhen-Yu Zhang 《World Journal of Gastroenterology》 SCIE CAS 2013年第23期3658-3664,共7页
AIM: To investigate the expression patterns of long non-coding RNAs (lncRNAs) in gastric cancer. METHODS: Two publicly available human exon arrays for gastric cancer and data for the corresponding normal tissue were d... AIM: To investigate the expression patterns of long non-coding RNAs (lncRNAs) in gastric cancer. METHODS: Two publicly available human exon arrays for gastric cancer and data for the corresponding normal tissue were downloaded from the Gene Expression Omnibus (GEO). We re-annotated the probes of the human exon arrays and retained the probes uniquely mapping to lncRNAs at the gene level. LncRNA expression profiles were generated by using robust multi-array average method in affymetrix power tools. The normalized data were then analyzed with a Bioconductor package linear models for microarray data and genes with adjusted P -values below 0.01 were considered differentially expressed. An independent data set was used to validate the results. RESULTS: With the computational pipeline established to re-annotate over 6.5 million probes of the Affymetrix Human Exon 1.0 ST array, we identified 136053 probes uniquely mapping to lncRNAs at the gene level. These probes correspond to 9294 lncRNAs, covering nearly 76% of the GENCODE lncRNA data set. By analyzing GSE27342 consisting of 80 paired gastric cancer and normal adjacent tissue samples, we identified 88 lncRNAs that were differentially expressed in gastric cancer, some of which have been reported to play a role in cancer, such as LINC00152, taurine upregulated 1, urothelial cancer associated 1, Pvt1 oncogene, small nucleolar RNA host gene 1 and LINC00261. In the validation data set GSE33335, 59% of these differentially expressed lncRNAs showed significant expression changes (adjusted P -value < 0.01) with the same direction. CONCLUSION: We identified a set of lncRNAs differentially expressed in gastric cancer, providing useful information for discovery of new biomarkers and therapeutic targets in gastric cancer. 展开更多
关键词 long NON-CODING RNA GASTRIC cancer MICROARRAY ANALYSIS data mining
暂未订购
融合联邦学习与LSTM模型的用户行为数据隐私保护机制研究
14
作者 程宇 龚亮华 《自动化与仪表》 2026年第1期134-138,共5页
近年来,用户行为数据在各类智能系统中广泛应用,其敏感性和分布式存储特性对隐私保护提出更高要求。为实现有效建模与隐私防护协同,研究构建融合联邦学习与LSTM模型的用户行为数据建模机制,并通过差分隐私与安全聚合机制提升整体安全性... 近年来,用户行为数据在各类智能系统中广泛应用,其敏感性和分布式存储特性对隐私保护提出更高要求。为实现有效建模与隐私防护协同,研究构建融合联邦学习与LSTM模型的用户行为数据建模机制,并通过差分隐私与安全聚合机制提升整体安全性。在模型训练过程中,上传参数经拉普拉斯机制扰动处理,同时引入SecAgg协议实现多方加密聚合。实验结果显示,模型预测准确率达到87.6%,通信成本控制在每轮约1.3 MB,训练收敛速度较传统联邦平均方法提升14.2%。综合评估表明,该机制在数据隐私保护、模型精度与系统通信效率之间实现了良好平衡。 展开更多
关键词 联邦学习 长短期记忆网络 用户行为建模 数据隐私保护 安全多方计算
在线阅读 下载PDF
多因素土壤墒情预测模型DA-LSTM-soil构建 被引量:2
15
作者 车银超 郑光 +3 位作者 熊淑萍 张明天 马新明 席磊 《河南农业大学学报》 北大核心 2025年第4期698-710,共13页
【目的】针对土壤墒情预测时特征因素复杂、预测精度不佳的问题,构建多因素土壤墒情预测模型DA-LSTM-soil,提高土壤墒情预测精度。【方法】以包含10个特征的气象和土壤时序数据作为输入,采用LSTM网络为基本单元,构建Encoder-Decoder网... 【目的】针对土壤墒情预测时特征因素复杂、预测精度不佳的问题,构建多因素土壤墒情预测模型DA-LSTM-soil,提高土壤墒情预测精度。【方法】以包含10个特征的气象和土壤时序数据作为输入,采用LSTM网络为基本单元,构建Encoder-Decoder网络结构,分别引入特征和时间两个注意力模块。利用河南省许昌市2020—2021年冬小麦生长过程中物联网监测站的气象、土壤数据集,对DA-LSTM-soil模型进行训练和测试。同时,利用DA-LSTM-soil模型对河南省4个不同土壤类型的小麦种植区的数据集进行预测。【结果】对比试验表明,相较于LSTM、CNN-LSTM、CNN-LSTM-attention、LSTM-attention等深度学习模型,DA-LSTM-soil模型在S_(RME)、S_(ME)、A_(ME)、R^(2)评价指标更优,分别达到0.1764、0.0311、0.0466、0.9938。消融试验显示,时间注意力对模型性能的提升高于特征注意力。对时间步的试验显示,用过往3000 min的数据进行预测时,模型性能最佳;模型精度随着预测时长的增加有所下降,然而在5000 min内,决定系数R2仍保持在0.7以上。【结论】利用注意力机制,DA-LSTMsoil模型在Encoder前计算不同气象和土壤因素对墒情影响的权重,在Decoder前计算数据的时序对墒情预测的权重,双阶段注意力机制在特征提取和权重分配方面的作用显著,使模型具有更好的预测性能和泛化能力,可以为田块尺度麦田土壤墒情预测提供技术依据。 展开更多
关键词 麦田 土壤墒情预测 时序数据 长短期记忆网络 注意力机制
在线阅读 下载PDF
The long-term trend of the sea surface wind speed and the wave height (wind wave, swell, mixed wave) in global ocean during the last 44 a 被引量:26
16
作者 ZHENG Chongwei ZHOU Lin +3 位作者 HUANG Chaofan SHI Yinglong LI Jiaxun LI Jing 《Acta Oceanologica Sinica》 SCIE CAS CSCD 2013年第10期1-4,共4页
Utilizing the 45 a European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis wave da- ta (ERA-40), the long-term trend of the sea surface wind speed and (wind wave, swell, mixed wave) wave height in ... Utilizing the 45 a European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis wave da- ta (ERA-40), the long-term trend of the sea surface wind speed and (wind wave, swell, mixed wave) wave height in the global ocean at grid point 1.5°× 1.5° during the last 44 a is analyzed. It is discovered that a ma- jority of global ocean swell wave height exhibits a significant linear increasing trend (2-8 cm/decade), the distribution of annual linear trend of the significant wave height (SWH) has good consistency with that of the swell wave height. The sea surface wind speed shows an annually linear increasing trend mainly con- centrated in the most waters of Southern Hemisphere westerlies, high latitude of the North Pacific, Indian Ocean north of 30°S, the waters near the western equatorial Pacific and low latitudes of the Atlantic waters, and the annually linear decreasing mainly in central and eastern equator of the Pacific, Juan. Fernandez Archipelago, the waters near South Georgia Island in the Atlantic waters. The linear variational distribution characteristic of the wind wave height is similar to that of the sea surface wind speed. Another find is that the swell is dominant in the mixed wave, the swell index in the central ocean is generally greater than that in the offshore, and the swell index in the eastern ocean coast is greater than that in the western ocean inshore, and in year-round hemisphere westerlies the swell index is relatively low. 展开更多
关键词 ECMWF reanalysis wave data wind wave SWELL mixed wave long-term trend swell index
在线阅读 下载PDF
面向长尾异构数据的个性化联邦学习框架 被引量:1
17
作者 吴家皋 易婧 +1 位作者 周泽辉 刘林峰 《计算机科学》 北大核心 2025年第9期232-240,共9页
针对数据长尾分布和异构性引起的联邦学习模型性能下降的问题,提出了一种新的个性化联邦学习框架——平衡的个性化联邦学习(Balanced Personalized Federated Learning,BPFed),将整个联邦学习过程分为基于个性化联邦学习的表示学习和基... 针对数据长尾分布和异构性引起的联邦学习模型性能下降的问题,提出了一种新的个性化联邦学习框架——平衡的个性化联邦学习(Balanced Personalized Federated Learning,BPFed),将整个联邦学习过程分为基于个性化联邦学习的表示学习和基于全局特征增强的个性化分类器再训练两个阶段。在第一阶段,首先采用Mixup策略进行数据增强,然后提出基于参数解耦的个性化联邦学习特征提取器训练方法,在优化特征提取器性能的同时减少通信开销;在第二阶段,首先提出新的基于全局协方差矩阵的类级特征增强方法,然后提出基于样本权重的标签平滑损失函数对客户端分类器进行平衡的个性化再训练,以纠正头类置信过度并提高尾类的泛化能力。大量的实验结果表明,在不同的数据长尾分布和异构性设置下,BPFed模型的准确度相比其他代表性相关算法均有明显提升。此外,消融和超参数影响实验也进一步验证了所提方法和优化策略的有效性。 展开更多
关键词 个性化联邦学习 长尾分布 数据异构性 参数解耦 特征增强 优化策略
在线阅读 下载PDF
基于CNN-BiLSTM的半监督学习地震波阻抗反演 被引量:1
18
作者 周萍 赵岩 《地球物理学进展》 北大核心 2025年第4期1812-1821,共10页
深度学习凭借其强大的特征提取能力,在各个领域展现出巨大潜力,为解决各种复杂问题提供了新的思路.深度学习模型往往需要大量的标记数据进行训练,但在实际应用中,由于成本原因,获得的测井数据有限,导致训练样本不足.为此,本文提出一种基... 深度学习凭借其强大的特征提取能力,在各个领域展现出巨大潜力,为解决各种复杂问题提供了新的思路.深度学习模型往往需要大量的标记数据进行训练,但在实际应用中,由于成本原因,获得的测井数据有限,导致训练样本不足.为此,本文提出一种基于CNN-BiLSTM的半监督学习地震波阻抗反演方法.采用内插重采样技术对波阻抗进行了增广,然后引入半监督学习策略对增广后数据进行训练,利用未标记数据信息来提高模型泛化能力和性能.Marmousi-2模型测试表明,仅用少量数据进行增广就能取得较好的反演效果,验证了该方法在小样本情况下是有效的. 展开更多
关键词 波阻抗反演 长短记忆神经网络 数据增广 半监督学习
原文传递
多源船载重力异常数据的联合再处理:以墨西哥湾为例 被引量:1
19
作者 郭金运 张鸿飞 +3 位作者 李真 祝程程 刘新 罗洪鑫 《武汉大学学报(信息科学版)》 北大核心 2025年第7期1277-1290,共14页
船载重力异常数据对于海洋重力场模型的构建起着至关重要的作用。虽然现有的船载重力异常数据在当初获取时已经经过重力测量常规改正、测量环境动态效应改正等处理,但是船载重力异常数据跨度时间较长,测量机构、所使用的重力仪与参考基... 船载重力异常数据对于海洋重力场模型的构建起着至关重要的作用。虽然现有的船载重力异常数据在当初获取时已经经过重力测量常规改正、测量环境动态效应改正等处理,但是船载重力异常数据跨度时间较长,测量机构、所使用的重力仪与参考基准均不尽相同,体现出一种多来源的特性;同时,受当时测量技术与数据处理技术的限制,获得的数据质量参差不齐,含有较大的误差,很难被直接应用。针对此问题,构建了一种多来源船载重力异常数据的联合再处理方法,分别从粗差剔除、长波误差改正、交叉点平差与系统误差补偿等方面对航线进行精化处理。以墨西哥湾为例,在美国国家环境信息中心获取该区域的船载重力异常数据,首先通过对比参考重力场对粗差航线及剩余航线粗差点进行剔除;其次,基于正常重力公式推导的二次多项式模型对测线长波误差进行改正;最后基于交叉点不符值,通过交叉点条件平差与构建混合多项式模型相结合的方法对航线观测值进行误差推估与系统误差补偿。经过一系列联合处理方法之后,船载重力异常交叉点不符值的均方根由原始数据的12.10 mGal减少至3.70 mGal,与重力异常模型SIO V32.1的残差均方根由精化前的6.70 mGal减少至3.91 mGal。对两组航线数据与模型的差值做频率域分析,结果表明误差的功率谱密度在各频率域上都得到明显减弱。以上结果均表明经过所提方法处理之后,船载数据质量得到了明显改善,该方法可用于进一步优化全球船载重力异常数据,从而为高精度的海洋重力场模型构建提供可靠数据集。 展开更多
关键词 船载重力测量 重力异常 墨西哥湾 数据再处理 长波误差改正 交叉点平差 功率谱密度
原文传递
国有股东参股如何降低民营企业股价同步性 被引量:1
20
作者 赵彦锋 《财经问题研究》 北大核心 2025年第9期46-61,共16页
股价同步性影响了资本市场价格信号的资源配置功能,而国有股东参股是否能够降低民营企业股价同步性值得关注。本文以2007—2023年我国沪深A股制造业民营上市公司为样本,考察国有股东参股对民营企业股价同步性的影响。研究发现:国有股东... 股价同步性影响了资本市场价格信号的资源配置功能,而国有股东参股是否能够降低民营企业股价同步性值得关注。本文以2007—2023年我国沪深A股制造业民营上市公司为样本,考察国有股东参股对民营企业股价同步性的影响。研究发现:国有股东参股能够显著降低民营企业股价同步性。异质性分析结果表明,从参股情况看,与单一国有股东相比,多个国有股东参股对民营企业股价同步性的降低效果更明显,国有股东向民营企业委派高管能够强化国有股东参股对民营企业股价同步性的降低效果;从企业特征看,国有股东参股对民营企业股价同步性的降低效果在年报信息可读性较高、内部控制质量较高的企业中更明显;从外部监督环境看,国有股东参股对民营企业股价同步性的降低效果在分析师关注度较高、网络媒体关注度较高的企业中更明显。机制分析结果表明,国有股东参股通过长期投资导向机制和市场关注机制降低民营企业股价同步性。本文的研究结论丰富了国有资本参股经济后果和股价同步性影响因素的研究,同时为民营企业利用国有资本发挥资本市场价值发现功能提供了理论与经验依据。 展开更多
关键词 国有股东参股 股价同步性 长期投资导向 数据资产 市场关注
在线阅读 下载PDF
上一页 1 2 68 下一页 到第
使用帮助 返回顶部