检索结果-南通市图书馆

Improved Keystroke Authentication Accuracy Based on Statistics and Weight

在线全文

学校读者我要写书评

暂无评论

China Communications 2012年第7期9卷 36-41页

作者： li Jian Guo Xiaojing li Meiyun li ruifan School of Computer Beijing University of Posts and TelecommunicationsBeijing 100876P.R.China

In order to improve the recognition accuracy of keystroke authentication, a methodology based on feature extraction of keystroke sequence is presented in this paper. Firstly, the data of the users' keystroke feature information that has too much deviation with the mean deviation is filtered out. Secondly, the probability of each input key is calculated and 10 values which do not have the best features are selected. Thirdly, they are weighed and a score evaluating the extent to which the user could be authenticated successfully is calculated. The benefit of using a third-party data set is more objective and comparable. At last, the experimental result shows that the detector in this paper is more accurate than the other three.

关键词：识别精度认证按键基础重量统计平均偏差特征提取

在线全文

学校读者我要写书评

暂无评论

增强提示学习的少样本文本分类方法

北京大学学报（自然科学版） 2024年第1期60卷 1-12页

作者：李睿凡魏志宇范元涛叶书勤张光卫北京邮电大学人工智能学院北京100876 教育部信息网络工程研究中心北京100876 交互技术与体验系统文化和旅游部重点实验室北京100876 北京邮电大学计算机学院北京100876

针对少样本文本分类任务,提出提示学习增强的分类算法(EPL4FTC)。该算法将文本分类任务转换成基于自然语言推理的提示学习形式,在利用预训练语言模型先验知识的基础上实现隐式数据增强,并通过两种粒度的损失进行优化。为捕获下游任务中... 详细信息

针对少样本文本分类任务,提出提示学习增强的分类算法(EPL4FTC)。该算法将文本分类任务转换成基于自然语言推理的提示学习形式,在利用预训练语言模型先验知识的基础上实现隐式数据增强,并通过两种粒度的损失进行优化。为捕获下游任务中含有的类别信息,采用三元组损失联合优化方法,并引入掩码语言模型任务作为正则项,提升模型的泛化能力。在公开的4个中文文本和3个英文文本分类数据集上进行实验评估,结果表明EPL4FTC方法的准确度明显优于所对比的基线方法。

关键词：预训练语言模型少样本学习文本分类提示学习三元组损失

C反应蛋白与白蛋白比值对川崎病早期丙种球蛋白抵抗的预测价值

在线全文

学校读者我要写书评

暂无评论

成都医学院学报 2024年第3期19卷 446-450页

作者：尹欢李蕊繁黄莉李静鲁利群宋悦阳长强成都医学院第一附属医院儿科成都610500 成都医学院第一附属医院心血管科成都610500

目的探讨C反应蛋白与白蛋白比值(CRP/ALB)对川崎病(KD)早期丙种球蛋白(IVIG)抵抗的预测价值。方法将2020年1月至2023年6月成都医学院第一附属医院住院的118例病程≤5 d的KD患儿分为IVIG抵抗组和IVIG敏感组,分析患儿临床资料,采用多因... 详细信息

目的探讨C反应蛋白与白蛋白比值(CRP/ALB)对川崎病(KD)早期丙种球蛋白(IVIG)抵抗的预测价值。方法将2020年1月至2023年6月成都医学院第一附属医院住院的118例病程≤5 d的KD患儿分为IVIG抵抗组和IVIG敏感组,分析患儿临床资料,采用多因素Logistic回归分析探究独立危险因素,ROC曲线探究CRP/ALB对KD患儿早期IVIG抵抗的预测价值。结果 IVIG抵抗组比IVIG敏感组有更高水平的白细胞计数(WBC)、CRP、CRP/ALB和更低水平的血红蛋白(HGB)。CRP/ALB升高和HGB降低是KD患儿早期IVIG抵抗的独立危险因素。CRP/ALB可预测KD患儿早期IVIG抵抗(AUC=0.664)。结论 CRP/ALB升高和HGB降低是KD患儿早期IVIG抵抗的独立危险因素,CRP/ALB可有效预测KD患儿早期IVIG抵抗。

关键词：川崎病 C反应蛋白与白蛋白比值丙种球蛋白抵抗

Fog Computing Architecture-Based Data Acquisition for WSN Applications

在线全文

学校读者我要写书评

暂无评论

China Communications 2017年第11期14卷 69-81页

作者： Guangwei Zhang ruifan li Institute of Network Technology Beijing University of Posts and Telecommunications 100876 China Engineering Research Center of Information Networks Ministry of Education 100876 China School of Computer Science Beijing University of Posts and Telecommunications 100876 China

Efficient and effective data acquisition is of theoretical and practical importance in WSN applications because data measured and collected by WSN is often unreliable, such as those often accompanied by noise and error, missing values or inconsistent data. Motivated by fog computing, which focuses on how to effectively offload computation-intensive tasks from resource-constrained devices, this paper proposes a simple but yet effective data acquisition approach with the ability of filtering abnormal data and meeting the real-time requirement. Our method uses a cooperation mechanism by leveraging on both an architectural and algorithmic approach. Firstly, the sensor node with the limited computing resource only accomplishes detecting and marking the suspicious data using a light weight algorithm. Secondly, the cluster head evaluates suspicious data by referring to the data from the other sensor nodes in the same cluster and discard the abnormal data directly. Thirdly, the sink node fills up the discarded data with an approximate value using nearest neighbor data supplement method. Through the architecture, each node only consumes a few computational resources and distributes the heavily computing load to several nodes. Simulation results show that our data acquisition method is effective considering the real-time outlier filtering and the computing overhead.

关键词： WSN fog computing abnormal data data filtering intrusion tolerance

High donor-number and low content electrolyte additive for stabilizing zinc metal anode

在线全文

学校读者我要写书评

暂无评论

Journal of Energy Chemistry 2024年第8期95卷 626-635页

作者： Yuxin Gong ruifan lin Bo Wang Huaizheng Ren Lei Wang Han Zhang Jianxin Wang Deyu li Yueping Xiong Dianlong Wang Huakun liu Shixue Dou MIIT Key Laboratory of Critical Materials Technology for New Energy Conversion and Storage State Key Laboratory of Urban Water Resource and Environment School of Chemistry and Chemical Engineering Harbin Institute of Technology Key Laboratory of Advanced Energy Materials Chemistry(Ministry of Education) College of Chemistry Nankai University Institute of Energy Material Science University of Shanghai for Science and Technology

The aqueous zinc ion batteries（AZIBs） are thought as promising competitors for electrochemical energy storage,though their wide application is curbed by the uncontrollable dendrite growth and gas evolution side reactions.Herein,to stabilize both zinc anodes and water molecules,we developed a modified electrolyte by adding a trace amount of N,N-diethylformanmide（DEF） into the ZnSO4electrolyte for the first time in zinc ion batteries.The effectiveness of DEF is predicted by the comparison of donor number and its preferential adsorption behavior on the zinc anode is further demonstrated by several spectroscopy characterizations,electrochemical methods,and molecular dynamics simulation.The modified electrolyte with 5% v.t.DEF content can ensure a stable cycling life longer than 3400 h of Zn‖Zn symmetric cells and an ultra-reversible Zn stripping/plating process with a high coulombic efficiency of 99.7%.The Zn‖VO2full cell maintains a capacity retention of 83.5% and a 104 mA h g-1mass capacity after 1000cycles.This work provides insights into the role of interfacial adsorption behavior and the donor number of additive molecules in designing low-content and effective aqueous electrolytes.

关键词： Aqueous zinc ion batteries Donor number Electrolyte additives Zinc anode Zinc dendrites

Offline Urdu Nastaleeq Optical Character Recognition Based on Stacked Denoising Autoencoder

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

China Communications 2017年第1期14卷 146-157页

作者： Ibrar Ahmad Xiaojie Wang ruifan li Shahid Rasheed Center for Intelligence of Science and Technology (CIST) School of Computer ScienceBeijing University of Posts and Telecommunications No. 10 Xitucheng Road Beijing 100876 China Department of Computer Science University of Peshawar Peshawar 25120 Pakistan Pakistan Telecommunication Company limited (PTCL) Islamabad 44000 Pakistan

Offline Urdu Nastaleeq text recognition has long been a serious problem due to its very cursive nature. In order to get rid of the character segmentation problems, many researchers are shifting focus towards segmentation free ligature based recognition approaches. Majority of the prevalent ligature based recognition systems heavily rely on hand-engineered feature extraction techniques. However, such techniques are more error prone and may often lead to a loss of useful information that might hardly be captured later by any manual features. Most of the prevalent Urdu Nastaleeq test recognition was trained and tested on small sets. This paper proposes the use of stacked denoising autoencoder for automatic feature extraction directly from raw pixel values of ligature images. Such deep learning networks have not been applied for the recognition of Urdu text thus far. Different stacked denoising autoencoders have been trained on 178573 ligatures with 3732 classes from un-degraded(noise free) UPTI(Urdu Printed Text Image) data set. Subsequently, trained networks are validated and tested on degraded versions of UPTI data set. The experimental results demonstrate accuracies in range of 93% to 96% which are better than the existing Urdu OCR systems for such large dataset of ligatures.

关键词： offline printed ligature recognition urdu nastaleeq denoising autoencoder deep learning classification

在线全文

学校读者我要写书评

暂无评论

飞秒激光永久光存储的发展及挑战

中国激光 2022年第10期49卷 152-168页

作者：王磊张栩王熠王琳王晴臧雪李瑞凡胥亦实李臻赜陈岐岱吉林大学电子科学与工程学院集成光电子学国家重点实验室吉林长春130012 清华大学精密仪器系精密测试技术与仪器国家重点实验室北京100084 吉林大学材料科学与工程学院吉林长春130012

面向大数据存储,总结当前冷数据存储的主要方式及特点,针对长寿命和高容量的需求,介绍飞秒激光永久光存储的概念和基本存储内涵;围绕透明介质材料体内改性的类型,依次介绍三维光存储和五维光存储的历史发展过程;阐述了当前具有双折射特... 详细信息

面向大数据存储,总结当前冷数据存储的主要方式及特点,针对长寿命和高容量的需求,介绍飞秒激光永久光存储的概念和基本存储内涵;围绕透明介质材料体内改性的类型,依次介绍三维光存储和五维光存储的历史发展过程;阐述了当前具有双折射特性的存储单元形成机制,超百层的高密度存储技术,225 kB/s单通道、潜在MB/s多通道的快速直写机制;并从纳米区域的电场连续性边界条件和光学衍射极限出发,展望飞秒激光永久光存储在存储容量和写入速度方面的挑战。

关键词：光电器件加工飞秒激光激光材料加工光与物质相互作用光存储纳米光栅

在线全文

学校读者我要写书评

暂无评论

基于无监督方法的电力文本专业词汇识别研究

电力工程技术 2020年第6期39卷 159-165页

作者：朱婷婷杜一帆李睿凡熊永平北京邮电大学人工智能学院北京100876 教育部信息网络工程研究中心北京100876 北京邮电大学计算机学院北京100876

电力专业词汇识别是面向变电运检文档进行深入语言理解和知识图谱构建等智能应用的基础。领域无关识别方法的效果不能令人满意,为此文中根据电力领域词汇的语言学特征提出一种面向电力领域的无监督专业词汇发现方法。首先以通用词典对... 详细信息

电力专业词汇识别是面向变电运检文档进行深入语言理解和知识图谱构建等智能应用的基础。领域无关识别方法的效果不能令人满意,为此文中根据电力领域词汇的语言学特征提出一种面向电力领域的无监督专业词汇发现方法。首先以通用词典对电力文档语料分词,然后根据电力专业词汇的特征设置不同大小的滑动窗口,将之前分词结果的多种组合作为候选词;进一步计算邻接变化度、信息熵、点态互信息以及词频等4种候选词统计量;最后基于综合语言学特征和成词边界3种语法规则对候选词进行筛选形成专业电力新词。在公开数据集上与基线方法进行了对比实验,实验结果验证了文中提出方法的有效性。

关键词：领域词典无监督学习新词识别滑动窗口统计特征

南方典型红壤侵蚀区马尾松林立木生物量无人机遥感估测

在线全文

学校读者我要写书评

暂无评论

西南林业大学学报（自然科学） 2024年第1期44卷 116-124页

作者：田上峰刘健余坤勇王瑞璠赵文凯福建农林大学林学院福建福州350002 福建农林大学3S技术与资源优化利用福建省高等学校重点实验室福建福州350002

以南方典型红壤侵蚀区长汀县河田镇为例,结合无人机与激光雷达产生的点云数据优势,通过局部最大值和分水岭算法反演单木树高(H)和冠层半径(R_(c)),拟合以H和R_(c)为变量组合的异速生长方程,得到以新冠层参数为底的马尾松立木生物量模型... 详细信息

以南方典型红壤侵蚀区长汀县河田镇为例,结合无人机与激光雷达产生的点云数据优势,通过局部最大值和分水岭算法反演单木树高(H)和冠层半径(R_(c)),拟合以H和R_(c)为变量组合的异速生长方程,得到以新冠层参数为底的马尾松立木生物量模型。结果表明:提取树高的决定系数(R^(2))和均方根误差(RMSE)分别为0.93和0.49 m;计算冠层半径的R2和RMSE分别为0.88和0.64 m;估算立木生物量的R^(2)和RMSE分别为0.89和3.37 kg。本研究通过无人机遥感影像定量参数并构建的异速生长方程中,以组合(H+R_(c))为底的异速生长方程估测马尾松林立木生物量的精度较高,可以有效估测马尾松林立木生物量,可为南方典型红壤侵蚀区马尾松林立木生物量准确估测提供参考。

关键词：马尾松红壤侵蚀区立木生物量生长方程冠层半径树高