期刊文献+
共找到411篇文章
< 1 2 21 >
每页显示 20 50 100
Structured Light 3D Vision Based on Laser Scanning Space-encoding and Its Synchro Control 被引量:1
1
作者 于晓洋 强锡富 +1 位作者 张健 吴丽莹 《Journal of Harbin Institute of Technology(New Series)》 EI CAS 1997年第2期20-24,共5页
This paper presents the principle and mathematic model for the 3D depth map method based on space encoding images performed by modulating scanning structuredlight according to time sequences,and the synchro control ... This paper presents the principle and mathematic model for the 3D depth map method based on space encoding images performed by modulating scanning structuredlight according to time sequences,and the synchro control among the camera,laser diode modulation and scanning polyhedron. 展开更多
关键词 D VISION structurED LIGHT space encoding LASER SCANNING SYNCHRO control
在线阅读 下载PDF
New Encoder Based on Grating Eddy-Current with Differential Structure
2
作者 ZHANG Zaigi LüNa +1 位作者 TAO Wei ZHAO Hui 《Journal of Shanghai Jiaotong university(Science)》 2025年第2期337-351,共15页
In response to the shortcomings of the common encoders in the industry,of which the photoelectric encoders have a poor anti-interference ability in harsh industrial environments with water,oil,dust,or strong vibration... In response to the shortcomings of the common encoders in the industry,of which the photoelectric encoders have a poor anti-interference ability in harsh industrial environments with water,oil,dust,or strong vibrations and the magnetic encoders are too sensitive to magnetic field density,this paper designs a new differential encoder based on the grating eddy-current measurement principle,abbreviated as differential grating eddy-current encoder(DGECE).The grating eddy-current of DGECE consists of a circular array of trapezoidal reflection conductors and 16 trapezoidal coils with a special structure to form a differential relationship,which are respectively located on the code plate and the readout plate designed by a printed circuit board.The differential structure of DGECE corrects the common mode interference and the amplitude distortion due to the assembly to some extent,possesses a certain anti-interference capability,and greatly simplifies the regularization algorithm of the original data.By means of the corresponding readout circuit and demodulation algorithm,the DGECE can convert the periodic impedance variation of 16 coils into an angular output within the 360°cycle.Due to its simple manufacturing process and certain interference immunity,DGECE is easy to be integrated and mass-produced as well as applicable in the industrial spindles,especially in robot joints.This paper presents the measurement principle,implementation methods,and results of the experiment of the DGECE.The experimental results show that the accuracy of the DGECE can reach 0.237%and the measurement standard deviation can reach±0.14°within360°cycle. 展开更多
关键词 encodER grating eddy-current differential structure angle measurement
原文传递
Application of deep autoencoder model for structural condition monitoring
3
作者 PATHIRAGE Chathurdara Sri Nadith LI Jun +2 位作者 LI Ling HAO Hong LIU Wanquan 《Journal of Systems Engineering and Electronics》 SCIE EI CSCD 2018年第4期873-880,共8页
Damage detection in structures is performed via vibra-tion based structural identification. Modal information, such as fre-quencies and mode shapes, are widely used for structural dama-ge detection to indicate the hea... Damage detection in structures is performed via vibra-tion based structural identification. Modal information, such as fre-quencies and mode shapes, are widely used for structural dama-ge detection to indicate the health conditions of civil structures.The deep learning algorithm that works on a multiple layer neuralnetwork model termed as deep autoencoder is proposed to learnthe relationship between the modal information and structural stiff-ness parameters. This is achieved via dimension reduction of themodal information feature and a non-linear regression against thestructural stiffness parameters. Numerical tests on a symmetri-cal steel frame model are conducted to generate the data for thetraining and validation, and to demonstrate the efficiency of theproposed approach for vibration based structural damage detec-tion. 展开更多
关键词 auto encoder non-linear regression deep auto en-coder model damage identification VIBRATION structural health monitoring
在线阅读 下载PDF
Dissemination and Genetic Structure of Carbapenemase Encoding Genes (bla<sub>OXA-23</sub>and bla<sub>OXA-24</sub>) in <i>Acinetobacter baumannii</i>from Southern Texas 被引量:1
4
作者 Nidha Azam Tamanna Talukder +1 位作者 Kava R. Robinson Dong H. Kwon 《Advances in Microbiology》 2015年第6期457-468,共12页
Acinetobacter baumannii is one of the most important human pathogens causing a variety of nosocomial infections. Carbapenem antibiotics have been primarily used to treat the A. baumannii infections. However, carbapene... Acinetobacter baumannii is one of the most important human pathogens causing a variety of nosocomial infections. Carbapenem antibiotics have been primarily used to treat the A. baumannii infections. However, carbapenem resistant A. baumannii producing carbapenemases causes serious treatment problems worldwide. Outbreaks of carbapenem resistant isolates have reported in some area of the United States, but their dissemination and genetic structure of the carbapenemase encoding genes are currently little known. To understand outbreaks, dissemination, and genetic structure of the carbapenemase encoding genes in Southern Texas, 32 clinical isolates collected from Austin and Houston, TX were characterized. Twenty-eight of 32 isolates were resistant to all tested β-lactam antibiotics including carbapenem (imipenem and meropenem). Three of them carried blaOXA-23 as a part of Tn2008 integrated into a known plasmid (pACICU2) and all others carried blaOXA-24 flanked by XerC/XerD-like recombinase binding sites that were adjoined by DNA sequences originated from multiple plasmids. Genotype analysis revealed that the 25 isolates carrying blaOXA-24 were all identical genotypes same as a representative isolate carrying blaOXA-24 from Chicago, IL but the 3 isolates carrying blaOXA-23 was a distinct genotype as compared with isolates carrying blaOXA-23 from Chicago, IL and Washington, D.C. Each of the blaOXA-23 and blaOXA-24 was transferred to carbapenem susceptible A. baumannii and E. coli with similar minimal inhibitory concentration (MIC) of carbapenem as that of their parental isolates but significantly lower levels of MIC in E. coli. Overall results suggest that a unique strain carrying blaOXA-23 and a similar strain carrying blaOXA-24 as seen in other geographic areas are currently disseminated in Southern Texas. 展开更多
关键词 Acinetobacter BAUMANNII DISSEMINATION and Genetic structure of Carbapenemase-encoding Genes
暂未订购
Estimation and Prediction of Gas Chromatography Retention Indices of Hydrocarbons in Straight-run Gasoline by Using Artificial Neural Network and Structural Coding Method
5
作者 YIN Chun sheng GUO Wei min +2 位作者 LIU Wei ZHAO Wei PAN Zhong xiao 《Chemical Research in Chinese Universities》 SCIE CAS CSCD 2001年第1期31-40,共10页
The molecular structures of hydrocarbons in straight run gasoline were numerically coded. The nonlinear quantitative relationship(QSRR) between gas chromatography(GC) retention indices of the hydrocarbons and their m... The molecular structures of hydrocarbons in straight run gasoline were numerically coded. The nonlinear quantitative relationship(QSRR) between gas chromatography(GC) retention indices of the hydrocarbons and their molecular structures were established by using an error back propagation(BP) algorithm. The GC retention indices of 150 hydrocarbons were then predicted by removing 15 compounds(as a test set) and using the 135 remained molecules as a calibration set. Through this procedure, all the compounds in the whole data set were then predicted in groups of 15 compounds. The results obtained by BP with the correlation coefficient and the standard deviation 0 993 4 and 16 54, are satisfied. 展开更多
关键词 structural encoding GC retention index Neural network Error back propagation(BP)
在线阅读 下载PDF
Path-partitioned encoding supports wildcard-awareness twig queries
6
作者 徐小双 冯玉才 王锋 《Journal of Shanghai University(English Edition)》 2009年第5期363-374,共12页
Finding all occurrences of a twig query in an XML database is a core operation for efficient evaluation of XML queries. It is important to effiectively handle twig queries with wildcards. In this paper, a novel path-p... Finding all occurrences of a twig query in an XML database is a core operation for efficient evaluation of XML queries. It is important to effiectively handle twig queries with wildcards. In this paper, a novel path-partitioned encoding scheme is proposed for XML documents to capture paths of all elements, and a twig query is modeled as an XPattern extended from tree pattern. After definition, simplification, normalization, verification and initialization of the XPattern, both work sets and a join plan are generated. According to these measures, an effiective algorithm to answer for a twig query, called DMTwig, is designed without unnecessary elements and invalid structural joins. The algorithm can adaptively deal with twig queries with branch ([ ]), child edge (/), descendant edge (//), and wildcard (*) synthetically. We show that path-partitioned encoding scheme and XPattern guarantee the I/O and CPU optimality for twig queries. Experiments on representative data set indicate that the proposed solution performs significantly. 展开更多
关键词 XML tree pattern structural join encoding scheme twig query
在线阅读 下载PDF
HiSC:A Hybrid XML Index Composing Structure-Encoded with Cluster
7
作者 YANG Jincai ZHANG Lin 《Wuhan University Journal of Natural Sciences》 CAS 2007年第5期840-844,共5页
A new way of indexing and processing twig patterns in an XML documents is proposed in this paper. Every path in XML document can be transformed into a sequence of labels by Structure-Encoded that constructs a one-to-o... A new way of indexing and processing twig patterns in an XML documents is proposed in this paper. Every path in XML document can be transformed into a sequence of labels by Structure-Encoded that constructs a one-to-one correspondence between XML tree and sequence. Base on identifying characteristics of nodes in XML tree, the elements are classified and clustered. During query proceeding, the twig pattern is also transformed into its Structure-Encoded. By performing subsequence matching on the set of sequences in XML documents, all the occurrences of path in the XML documents are refined. Using the index, the numbers of elements retrieved are minimized. The search results with pertinent format provide more structure information without any false dismissals or false alarms. The index also supports keyword search Experiment results indicate the index has significantly efficiency with high precision. 展开更多
关键词 XML index structure cluster index region coding structure-encoded keyword search
在线阅读 下载PDF
An Improved Method of Linear Encoding Representation-A Cell Linear Quadtree
8
作者 He Yu Zhang Qianren Ye Yizheng Li Zhongrong (Department of Computer science) 《哈尔滨工业大学学报》 EI CAS CSCD 北大核心 1990年第3期28-31,共4页
The linear encoding of a quadtree is an efficient way to represent the quadtree. In this paper, an improved linear quadtree, a cell linear quadtree, is proposed, in which its redundant storage is eliminated and the co... The linear encoding of a quadtree is an efficient way to represent the quadtree. In this paper, an improved linear quadtree, a cell linear quadtree, is proposed, in which its redundant storage is eliminated and the concept of a cell is introduced. Therefore, it has higher storage efficiency than a conventional linear quadtree. 展开更多
关键词 分支树 图像处理 图像编码 层状数据结构 编码表达法
在线阅读 下载PDF
基于权重共享的轻量化注意力卷积编码器
9
作者 肖芷翊 奚峥皓 《现代电子技术》 北大核心 2026年第2期44-48,共5页
环境的复杂多样导致车载传感器获得的图像存在大量的干扰信息,给交通标志的识别带来了较大困难。为此,提出一种基于权重共享预训练的轻量化注意力卷积编码器。首先,结合注意力机制与非对称卷积结构,对已编码的重要特征进行关注;然后,利... 环境的复杂多样导致车载传感器获得的图像存在大量的干扰信息,给交通标志的识别带来了较大困难。为此,提出一种基于权重共享预训练的轻量化注意力卷积编码器。首先,结合注意力机制与非对称卷积结构,对已编码的重要特征进行关注;然后,利用对比学习中的并行结构对编码器进行预训练,提升编码器在复杂环境下的样本学习能力;最后,对输出的特征编码进行预测,完成对交通标志的识别。实验结果表明,所提算法在模糊和遮挡条件下的准确率分别达到94.65%、91.23%,具有实用性。 展开更多
关键词 通道注意力机制 卷积编码器 非对称卷积结构 权重共享 对比学习 图像识别
在线阅读 下载PDF
耦合Encoder-Decoder的LSTM径流预报模型研究 被引量:14
10
作者 林康聆 陈华 +3 位作者 陈清勇 罗宇轩 刘峰 陈杰 《武汉大学学报(工学版)》 CAS CSCD 北大核心 2022年第8期755-761,共7页
将长短期记忆神经网络(long short-term memory neural network,LSTM)与Encoder-Decoder结构耦合应用为LSTM-ED模型,并与LSTM人工智能径流预报模型进行比较。通过在闽江建溪流域进行应用,结果表明,相较于LSTM,LSTM-ED在检验期整体和各... 将长短期记忆神经网络(long short-term memory neural network,LSTM)与Encoder-Decoder结构耦合应用为LSTM-ED模型,并与LSTM人工智能径流预报模型进行比较。通过在闽江建溪流域进行应用,结果表明,相较于LSTM,LSTM-ED在检验期整体和各预见期具有更高的精度和稳定性,且对于典型洪水的预报洪峰误差更小,其独有的语义向量可以保持水文信息的连续性,预报径流过程不易受降雨波动干扰。2个模型的预报能力都与流域最大汇流时间密切相关,当预见期小于流域最大汇流时间时,2个模型都有很好的预报能力;当预见期大于流域最大汇流时间时,模型预报能力显著变差;当预见期远大于流域最大汇流时间时,2个模型都失去预报可靠性。 展开更多
关键词 径流预报 encoder-Decoder结构 长短期记忆神经网络 深度学习 人工神经网络
原文传递
Design and implementation of binary tree data structure based on DNA computing
11
作者 YANG Chun-de WEI Guo-hui 《重庆邮电大学学报(自然科学版)》 北大核心 2009年第2期292-295,共4页
The designing,encodings and an instance of simulation of a binary tree for DNA computer were proposed,which utilizes the method of biology to complete inserting and deleting of the binary tree. Firstly,DNA encodings f... The designing,encodings and an instance of simulation of a binary tree for DNA computer were proposed,which utilizes the method of biology to complete inserting and deleting of the binary tree. Firstly,DNA encodings for storage and all elements of the binary tree were completely given out. Then, the implementations of all bio-operations in DNA computer were described. Finally, to prove the feasibility of this method, an actual binary tree with detailed nucleotide encodings was introduced. The process of an algorithm implemented on this binary tree was demonstrated. Based on this method, more other data structures in DNA computer can be developed. 展开更多
关键词 DNA计算机 完全二叉树 数据结构 设计 生物学方法 基因编码 核苷酸 证明
在线阅读 下载PDF
Projecting Distortion Calibration and Evaluation of Coding Fringes in Structured Light System
12
作者 Haibin Wu Qing Xu +3 位作者 Xi Wu Guanglu Sun Xiaoyang Yu Xiaoming Sun 《国际计算机前沿大会会议论文集》 2016年第1期139-140,共2页
Coded structured light is an accurate, fast 3D measurement approach with high sampling density, of which the encoded fringes are distorted when projected to curved surface. Focused on the demand of encoding, decoding,... Coded structured light is an accurate, fast 3D measurement approach with high sampling density, of which the encoded fringes are distorted when projected to curved surface. Focused on the demand of encoding, decoding, multiview registration and system calibration, we expect to obtain undistorted fringes from camera image. Therefore, in this paper, we analyze the accuracy and sampling density of projecting distortion calibration approach based on control point and fitting surface. Moreover, combining the characteristic of coded structured light system, we design encoded fringe projecting distortion calibration scheme based on simplified encoded structured light model. Primarily, we neglect the minor parameters that affect the calibration in structured light model to reduce complexity. Then, we build the correspondence between camera image points and projector image points and achieve the calibration. Finally, we design evaluation scheme of projecting distortion calibration with parallelism and equal interval, and verify the effectiveness and accuracy of the approach through visual effect and experimental data. 展开更多
关键词 CODED structured light Projecting DISTORTION GEOMETRIC CALIBRATION encoded FRINGE
在线阅读 下载PDF
语体语法核心概念的阐析与推展 被引量:3
13
作者 冯胜利 《华文教学与研究》 2025年第2期77-89,共13页
语体语法自2010年提出后,得到学界广泛的关注和反响:有的直触要脉,促其深入;有的兼思别异,旁拓蹊径;也有的取解不同,而致歧其旨。凡此种种虽曰自然,亦激发藑思与辨析。譬如“语体何以为语法”乃结构之语距属性使然,而非语境使用所致;再... 语体语法自2010年提出后,得到学界广泛的关注和反响:有的直触要脉,促其深入;有的兼思别异,旁拓蹊径;也有的取解不同,而致歧其旨。凡此种种虽曰自然,亦激发藑思与辨析。譬如“语体何以为语法”乃结构之语距属性使然,而非语境使用所致;再如“形式-功能对应律”其实乃“形式-功能对生律”等等。本文就上述相关的核心概念进行了深入辨析与阐释,旨在厘清语体语法因“拉距”而“对立”、因对立而“两极”、因两极而成“结构”的概念要旨及其内涵与推展。据此,以往语言学(包括社会语言)中所谓“变异”形式便不再是“因用而现”的个别现象,而具有了因“拉距”而形成的“正式-非正式”的对立关系。文章继而从语体语法理论构建的角度追溯语体和语法关系研究从“对应”到“对生”再到“语体编码”的三阶段,提出语体语法学是对人类语言交际中固有[±正式]特征编码结构的系统解释。文章最后从“语法悖论”和“语法性”两个角度阐释语体语法的理论意义,及其“功能形式化”和生成“递归性”的理论要旨。 展开更多
关键词 语体语法 语体核心概念 语距 结构的语体编码 语法悖论
在线阅读 下载PDF
机器学习模型与物理机制模型在长诏水库流域实时洪水预报中的比较研究
14
作者 瞿思敏 余裕 +5 位作者 方正 罗小亮 石朋 虞鸿 张锏 李倩 《水资源保护》 北大核心 2025年第5期73-78,88,共7页
以曹娥江长诏水库流域为研究区域,选择影响洪水过程的降雨、径流作为主要影响因子构建流域长短期记忆网络(LSTM)模型,分析流域水文气象特征和产汇流机理,并与新安江模型模拟结果进行对比分析。结果表明:LSTM模型和新安江模型在长诏水库... 以曹娥江长诏水库流域为研究区域,选择影响洪水过程的降雨、径流作为主要影响因子构建流域长短期记忆网络(LSTM)模型,分析流域水文气象特征和产汇流机理,并与新安江模型模拟结果进行对比分析。结果表明:LSTM模型和新安江模型在长诏水库流域洪水模拟中应用效果较好,LSTM模型合格率更高,且LSTM模型平均径流深和洪峰模拟结果的相对误差更小,精度更高,而新安江模型确定性系数比较稳定且峰现时差更小;LSTM模型降低了对人为经验的依赖,可用于对精度要求较高的实时洪水预报;新安江模型对于一些突发事件能够结合参数表达的物理过程解释误差来源,更适用于极端洪水等复杂情景分析和物理过程解释的研究。 展开更多
关键词 长诏水库流域 洪水预报 新安江模型 LSTM模型 编码-解码结构
在线阅读 下载PDF
基于跨模态级联扩散模型的图像描述方法
15
作者 陈巧红 郭孟浩 +1 位作者 方贤 孙麒 《浙江大学学报(工学版)》 北大核心 2025年第4期787-794,共8页
现有文本扩散模型方法无法有效根据语义条件控制扩散过程,扩散模型训练过程的收敛较为困难,为此提出基于跨模态级联扩散模型的非自回归图像描述方法.引入跨模态语义对齐模块用于对齐视觉模态和文本模态之间的语义关系,将对齐后的语义特... 现有文本扩散模型方法无法有效根据语义条件控制扩散过程,扩散模型训练过程的收敛较为困难,为此提出基于跨模态级联扩散模型的非自回归图像描述方法.引入跨模态语义对齐模块用于对齐视觉模态和文本模态之间的语义关系,将对齐后的语义特征向量作为后续扩散模型的语义条件.通过设计级联式的扩散模型逐步引入丰富的语义信息,确保生成的图像描述贴近整体语境.增强文本扩散过程中的噪声计划以提升模型对文本信息的敏感性,充分训练模型以增强模型的整体性能.实验结果表明,所提方法能够生成比传统图像描述生成方法更准确和丰富的文本描述.所提方法在各项评价指标上均明显优于其他非自回归文本生成方法,展现了在图像描述任务中使用扩散模型的有效性和潜力. 展开更多
关键词 深度学习 图像描述 扩散模型 多模态编码器 级联结构
在线阅读 下载PDF
基于发动机罩板体素数据的头部碰撞损伤快速预测
16
作者 陈光 尚云龙 +2 位作者 张君媛 唐洪斌 孙秀秀 《汽车技术》 北大核心 2025年第10期52-62,共11页
为了快速评估车辆开发阶段不同车辆发动机罩板结构的行人保护性能,提出了一种基于三维体素数据的机器学习预测方法。首先,基于7辆车的有限元模型构建以罩板头部碰撞点为中心的三维体素数据集,精确表征罩板的复杂三维结构;其次,利用3D-CN... 为了快速评估车辆开发阶段不同车辆发动机罩板结构的行人保护性能,提出了一种基于三维体素数据的机器学习预测方法。首先,基于7辆车的有限元模型构建以罩板头部碰撞点为中心的三维体素数据集,精确表征罩板的复杂三维结构;其次,利用3D-CNN自编码器对三维体素数据进行无监督特征学习,生成1000维的编码数据集;随后,通过PCA线性降维技术过滤噪声特征,得到224维PCA数据集;最后,对比数据集在4种机器学习模型上的性能。试验结果表明:224维PCA数据集的预测效果更优,其中,SVR模型对HIC和a峰的预测决定系数R²分别达到98.86%和98.44%,显著提升了预测精度和泛化能力。此外,单辆车超过100组碰撞点的计算时间缩短至10 s内,大幅提高了汽车安全性能开发效率。 展开更多
关键词 行人保护 机器学习 头部损伤指标 罩板结构 3D-CNN 自编码器
在线阅读 下载PDF
基于双Transformer结构的多模态视频段落描述生成研究
17
作者 赵宏 张立军 《计算机工程与应用》 北大核心 2025年第21期182-191,共10页
针对现有视频段落描述方法对视频中主要事件的关注度不足与多事件描述之间缺乏连贯性的问题,在现有编码器-解码器框架的基础上,提出了一种基于双Transformer结构的多模态视频段落描述模型。采用Faster-RCNN对视频中心帧目标进行细粒度... 针对现有视频段落描述方法对视频中主要事件的关注度不足与多事件描述之间缺乏连贯性的问题,在现有编码器-解码器框架的基础上,提出了一种基于双Transformer结构的多模态视频段落描述模型。采用Faster-RCNN对视频中心帧目标进行细粒度特征提取,由混合注意力结合全局视觉特征选择最具代表性的细粒度局部视觉特征,对视频中主要事件信息进行补充与增强,提高视频内容描述的准确性;提出在Transformer结构中增加存储模块与混合注意力模块,并设计了双Transformer结构,内部Transformer对事件内一致性进行建模,外部Transformer由混合注意力计算与当前事件最相关的状态建模事件间的一致性,结合内外部Transformer的输出对事件内容进行预测,提高生成描述语句的连贯性。在ActivityNet Captions数据集和YouCookII数据集上的实验结果表明,所提模型在BLEU-4、METEOR、ROUGE-L和CIDEr指标上相较于现有主流视频段落描述模型有明显提升,验证了模型的有效性。 展开更多
关键词 视频段落描述 编码器-解码器结构 细粒度局部视觉特征 双Transformer结构
在线阅读 下载PDF
基于语义分割的静脉输液液位检测
18
作者 邹学玉 熊莉娅 +1 位作者 周箩鱼 喻旸 《长江大学学报(自然科学版)》 2025年第4期119-126,共8页
为了解决人工监控静脉输液过程效率低问题,利用摄像头拍摄的输液容器图片,对输液容器液位线进行检测。通过语义分割来预测输液图片中每个像素点的标签,将液位检测问题转化为三类分割问题:背景、有液区域和无液区域。其中,将背景分割出... 为了解决人工监控静脉输液过程效率低问题,利用摄像头拍摄的输液容器图片,对输液容器液位线进行检测。通过语义分割来预测输液图片中每个像素点的标签,将液位检测问题转化为三类分割问题:背景、有液区域和无液区域。其中,将背景分割出来获得输液目标,再以液位线为界限把输液目标分割为有液区域和无液区域,从而实现输液目标液位检测。首先,通过标注不同光照条件、容量和背景下的单目标和双目标图片,创建输液目标数据集。然后,搭建基于VGG16(visual geometry group)的U-Net模型和基于Mobilenetv2、ResNet50的金字塔场景解析网络(PSPNet)模型。三个网络结合迁移学习方法,弥补自建输液数据集不足问题,加速模型学习效率。试验结果表明,三个网络都能分割输液目标,其中,基于VGG16网络的U-Net模型更优,其平均交并比(MIoU)为89.5%,平均像素精度(MPA)为94.8%。 展开更多
关键词 液位检测 语义分割 编码解码结构 卷积神经网络 残差网络
在线阅读 下载PDF
基于BERT融合算法的病例文本结构化模型研究
19
作者 张雪 王琛琛 职宁 《中国医疗设备》 2025年第9期12-19,共8页
目的为提升临床病例文本中非结构化信息的提取效率与准确性,推动医学智能化发展,本研究提出一种基于双向编码器表示融合算法的病例文本结构化模型。方法该模型采用双向编码器进行语义表示,利用图卷积神经网络提取词语间局部依赖,融合长... 目的为提升临床病例文本中非结构化信息的提取效率与准确性,推动医学智能化发展,本研究提出一种基于双向编码器表示融合算法的病例文本结构化模型。方法该模型采用双向编码器进行语义表示,利用图卷积神经网络提取词语间局部依赖,融合长短时记忆网络建立时序关系,并引入条件随机场优化标签序列的一致性。实验选用MIMIC-Ⅲ和ClinicalSTS这2个权威临床数据集进行分析,构建五类医学文本分类任务,对比所有模型在不同结构组合下的性能表现。结果本研究所提出的病例文本结构化模型的精准度、召回率和F1得分分别为0.92、0.90和0.91,均较传统双向编码器模型提升约10%;在处理超过1000字的长文本场景时,模型效率提升达12%,表现出良好的时效性与可扩展性。结论本研究通过验证深度融合多种结构对提高病例文本结构化处理能力的有效性,为智能医学文本分析提供了理论依据。 展开更多
关键词 病例文本 长短时记忆网络 BERT 图卷积神经网络 结构化模型 医学智能化
在线阅读 下载PDF
编码器-解码器结构的刀具磨损状态预测研究 被引量:2
20
作者 刘本刚 吴文江 +2 位作者 赵丹 王裴岩 彭春杨 《小型微型计算机系统》 北大核心 2025年第6期1530-1536,共7页
针对航空钛合金加工中刀具磨损状态监测难题,提出了面向刀具磨损状态预测的编码器-解码器网络结构,构建了Transformer、BiLSTM、BiGRU等72种组合模型,通过在航空钛合金高效加工实测数据样本集上验证发现:以Transformer为编码器的模型性... 针对航空钛合金加工中刀具磨损状态监测难题,提出了面向刀具磨损状态预测的编码器-解码器网络结构,构建了Transformer、BiLSTM、BiGRU等72种组合模型,通过在航空钛合金高效加工实测数据样本集上验证发现:以Transformer为编码器的模型性能最优,其中Transformer-BiGRU组合模型F1值达69.61%,显著优于GS-XGBoost(58.01%)、Attention-CNN(57.65%)等方法,研究表明基于编码器-解码器的刀具状态预测模型在航空钛合金复杂切削工况下具有显著优势,未来可通过模型优化和扩充样本数据进一步提升其性能. 展开更多
关键词 编码器-解码器结构 刀具磨损状态预测 TRANSFORMER 双向循环神经网络 航空钛合金高效加工
在线阅读 下载PDF
上一页 1 2 21 下一页 到第
使用帮助 返回顶部