Feature combination via importance-inhibition analysis

A new method for combining features via importance-inhibition analysis （IIA） is described to obtain more effective feature combination in learning question classification. Features are combined based on the inhibition among features as well as the importance of individual features. Experimental results on the Chinese questions set show that, the IIA method shows a gradual increase in average and maximum accuracies at all feature combinations, and achieves great improvement over the importance analysis（IA） method on the whole. Moreover, the IIA method achieves the same highest accuracy as the one by the exhaustive method, and further improves the performance of question classification.

A Fast Image Matching Algorithm Using a Combination of Line Segment Features

The scene matching navigation is a research focus in the field of autonomous navigation,but the real-time performance of image matching algorithm is difficult to meet the needs of real navigation systems.Therefore,this paper proposes a fast image matching algorithm.The algorithm improves the traditional line segment extraction algorithm and combines with the Delaunay triangulation method.By combining the geometric features of points and lines,the image feature redundancy is reduced.Then,the error with confidence criterion is analyzed and the matching process is completed.The simulation results show that the proposed algorithm can still work within 3°rotation and small scale variation.In addition,the matching time is less than 0.5 s when the image size is 256 pixel×256 pixel.The proposed algorithm is suitable for autonomous navigation systems with multiple feature distribution and higher real-time requirements.

APPLICATION RESEARCH OF COMBINATION PRINCIPLE IN CAD FEATURE RECOGNIZATION

The method of extracting the basic features of part from the file of STEPAP214 of 3-D model is proposed. All faces in the file are the minimal elements. The combination isdone for the faces with geometry restrictions and some attributes by the theory of the bestalphabetic tree which is constructed by HU--TUCKER algorithm in combination principle. So the basicfeatures could be attained. This provides the research basis to the more share and integration ofCAD information in the virtual enterprises. Finally, a case is used to illustrate the validity ofthe approach.

A Sequence Image Matching Method Based on Improved High-Dimensional Combined Features

Image matching technology is theoretically significant and practically promising in the field of autonomous navigation.Addressing shortcomings of existing image matching navigation technologies,the concept of high-dimensional combined feature is presented based on sequence image matching navigation.To balance between the distribution of high-dimensional combined features and the shortcomings of the only use of geometric relations,we propose a method based on Delaunay triangulation to improve the feature,and add the regional characteristics of the features together with their geometric characteristics.Finally,k-nearest neighbor(KNN)algorithm is adopted to optimize searching process.Simulation results show that the matching can be realized at the rotation angle of-8°to 8°and the scale factor of 0.9 to 1.1,and when the image size is 160 pixel×160 pixel,the matching time is less than 0.5 s.Therefore,the proposed algorithm can substantially reduce computational complexity,improve the matching speed,and exhibit robustness to the rotation and scale changes.

Clinical features of multiple trauma patients combined with spine and spinal cord injuries

Objective To analyze the clinical features of the multiple trauma patients combined with spine and spinal cord injuries.Methods A retrospective study was performed in143multiple trauma patients combined with spine and spinal

Application of Multimedia Technology in Feature Wall Design

New value systems and new aesthetic experience have been formed in the new era,triggered by reform of technical means and transformation of social economy. New multimedia technologies will produce new feelings,new experience and new ways of thinking,which correspondingly result in changes of landscape designs. Virtual reality technology is used to build autokinetic effect and mobility of feature wall,complex structures and forms,vivid and creative landscape elements,so that the landscapes will interact with users intimately,and humanized landscape spaces will be created.

Modification of evidence theory based on feature extraction

Although evidence theory has been widely used in information fusion due to its effectiveness of uncertainty reasoning, the classical DS evidence theory involves counter-intuitive behaviors when high conflict information exists. Many modification methods have been developed which can be classified into the following two kinds of ideas, either modifying the combination rules or modifying the evidence sources. In order to make the modification more reasonable and more effective, this paper gives a thorough analysis of some typical existing modification methods firstly, and then extracts the intrinsic feature of the evidence sources by using evidence distance theory. Based on the extracted features, two modified plans of evidence theory according to the corresponding modification ideas have been proposed. The results of numerical examples prove the good performance of the plans when combining evidence sources with high conflict information.

Thinking and practice on clinical safety evaluation of combination of traditional Chinese medicine and western medicine

China is gradually establishing a multidisciplinary diagnosis and treatment system of traditional Chinese medicine(TCM)and western medicine.TCM-drug combination is prone to adverse reactions.Clinical feature is the appearance of adverse reactions,and target is the internal mechanism.The establishment of feature and target correlation model will contribute to the development of this field.This paper introduces the four steps of feature-target correlation method that risk identification,feature extraction,sign target correlation and experimental research.Xiyanping-Ribavirin combination is as an example to illustrate this method.It is expected that the method will be popularized and applied to protect clinical safety.

Research on Technology of Twin Image Recognition Based on the Multi-feature Fusion

In order to improve the accuracy and stability of fruit and vegetable image recognition by single feature,this project proposed multi-feature fusion algorithms and SVM classification algorithms.This project not only introduces the Reproducing Kernel Hilbert space to improve the multi-feature compatibility and improve multi-feature fusion algorithm,but also introduces TPS transformation model in SVM classifier to improve the classification accuracy,real-time and robustness of integration feature.By using multi-feature fusion algorithms and SVM classification algorithms,experimental results show that we can recognize the common fruit and vegetable images efficiently and accurately.

Web page classification based on heterogeneous features and a combination of multiple classifiers

Precise web page classification can be achieved by evaluating features of web pages, and the structural features of web pages are effective complements to their textual features. Various classifiers have different characteristics, and multiple classifiers can be combined to allow classifiers to complement one another. In this study, a web page classification method based on heterogeneous features and a combination of multiple classifiers is proposed. Different from computing the frequency of HTML tags, we exploit the tree-like structure of HTML tags to characterize the structural features of a web page. Heterogeneous textual features and the proposed tree-like structural features are converted into vectors and fused. Confidence is proposed here as a criterion to compare the classification results of different classifiers by calculating the classification accuracy of a set of samples. Multiple classifiers are combined based on confidence with different decision strategies, such as voting, confidence comparison, and direct output, to give the final classification results. Experimental results demonstrate that on the Amazon dataset, 7-web-genres dataset, and DMOZ dataset, the accuracies are increased to 94.2%, 95.4%, and 95.7%, respectively. The fusion of the textual features with the proposed structural features is a comprehensive approach, and the accuracy is higher than that when using only textual features. At the same time, the accuracy of the web page classification is improved by combining multiple classifiers, and is higher than those of the related web page classification algorithms.

Subdivision surface modeling system based on arbitrary topological curves network and combined subdivision

Arbitrary topological curve network has no restriction in topology structure,so it has more powerful representing ability in defining complex surfaces.A complex surface modeling system is presented based on arbitrary topological curve network and the improved combined subdivision method,its functions including creating and editing curve network,and generating and modifying curve network＇s interpolated surface.This modeling system can be used to the process of products＇concept design,and its applications is also significant to the development of subdivision method.

A Lightweight Improved U-Net with Shallow Features Combination and Its Application to Defect Detection

In order to solve the problems of shallow features loss and high computation cost of U-Net,we propose a lightweight with shallow features combination(IU-Net).IU-Net adds several convolution layers and short links to the skip path to extract more shallow features.At the same time,the original convolution is replaced by the depth-wise separable convolution to reduce the calculation cost and the number of parameters.IU-Net is applied to detecting small metal industrial products defects.It is evaluated on our own SUES-Washer dataset to verify the effectiveness.Experimental results demonstrate that our proposed method outperforms the original U-Net,and it has 1.73%,2.08%and 11.2%improvement in the intersection over union,accuracy,and detection time,respectively,which satisfies the requirements of industrial detection.

顾及特征离散程度的SEaTH特征优化选择方法

特征选择是面向对象信息提取的关键步骤之一。本文针对分离阈值(separability and threshold,SEaTH)这一特征选择方法在实际应用中存在的不足,例如未考虑特征值的离散程度,仅利用J-M距离评判单一特征,特征间可能存在较强相关性,以及无法有效确定出分类顺序,提出了一种改进的SEaTH算法(optimized SEaTH,OPSEaTH)。OPSEaTH算法首先在J-M距离基础上构建了一类特征评价指标(E值),有效解决了特征值的离散度问题;然后,基于E值构建出特征组合评价指标(C_(e)值),可有效评估得到每种地物的最佳特征组合并自动确定出地物的分类顺序;最后基于eCognition等分类器可完成对地物对象的最终有效分类。利用高分二号遥感影像数据对本文方法进行了测试,并将结果分别与SEaTH算法、DPC、OIF和最近邻分类器的分类结果进行了对比,结果表明:OPSEaTH算法不仅能有效降低特征维数、优化特征空间,还能够对分类顺序进行自动化合理确定,总体精度和Kappa系数及其他精度指标,均显著优于基于SEaTH算法的特征选择结果。本文方法无论从特征降维效果、分类结果精度还是计算效率方面均优于DPC、OIF和最近邻分类器结果。OPSEaTH是一种更优的特征选择方法。

组合光透视下小面积室内封闭空间分割方法

为提高室内空间的优化规划设计能力,提出基于点云数据语义分割的组合光透视下小面积室内封闭空间分割方法。构建小面积室内封闭空间的环境三维信息感知模型,采用室内空间点云提取室内封闭空间图像的坐标信息,结合语义组合特征分割方法将融合空间信息映射到高分辨空间异质单元结构中,引入几何约束的子空间投影特征信息,结合组合光透视的高分辨率分割图模型参数融合方法,提取小面积室内封闭的端元,采用点云数据语义分割方法实现空间分割。结果表明,采用该方法有效实现对复杂室内场景的三维重建,空间分割的均方根误差较低为0.808%,峰值信噪比较高42.156 dB,空间分割速度较快,平均为12.83 ms。

基于病证结合特点的绝经后骨质疏松症动物模型分析

绝经后骨质疏松症(PMOP)病证结合动物模型在中医药研究中应用较多,但建立方法各式各样且缺乏统一的评价标准。文章从中医学和现代医学临床角度出发,就目前PMOP模型动物选择、构建方法和模型动物与临床吻合度进行对比,结合病证结合动物模型构建思路,从PMOP病证结合动物模型的构建和进一步完善评价方法进行归纳总结,以期为PMOP实验研究提供新思路。

试验环境水下声信号的特征提取方法

水下试验环境参数的反演是水声学研究领域的重要内容。而当前研究的关键是通过对水下声信号做特征提取从而获取参数信息。针对特征提取较难、模型很难拟合等问题。本文提出了一种试验环境水下声信号的特征提取方法。将水下声信号同时用梅尔频谱倒谱系数及线性预测系数处理,两者运用特征加权组合方法得到新的特征矩阵;再应用映射插值算法对特征矩阵进行处理,获得适应神经网络输入的三通道矩阵。本文选取的网络模型为残差神经网络。利用实验室所录制的对河口水库数据集测试表明,本文提出的特征提取方法普遍优于仅利用梅尔频谱倒谱系数或线性预测系数的特征处理方法。利用单频矩形脉冲信号对环境进行深度5分类,准确率平均提升2%。利用线性调频信号对环境进行深度5分类,准确率平均提升2.03%。本文提出的特征提取方法对线性调频信号在深度分类任务下处理的结果要优于单频矩形脉冲信号处理的结果。

考虑时序特征的深圳港集装箱吞吐量组合方法预测

集装箱吞吐量预测对港口企业运营及决策具有重要的作用。传统集装箱吞吐量预测方法存在预测精度不高的缺点。为解决这一问题,提出了一种考虑季节性和不确定性的SARIMA-XGBoost组合预测方法。针对集装箱吞吐量的季节性特征,选取季节性自回归移动平均模型(seasonal autoregressive integrated moving average model,SARIMA)捕捉周期性特征和线性特征;针对集装箱吞吐量中的不确定性因素,选取极致梯度提升树算法(extreme gradient boosting,XGBoost)自适应学习时间序列数据中的复杂模式和非线性特征。通过选取优化指标并计算分配权重的方式实现了预测模型中线性和非线性特征的有效融合,从而提升预测精度。通过对深圳港2013—2022年集装箱吞吐量月度数据进行实证研究和对比分析,结果表明SARIMA-XGBoost组合方法预测精度最高、稳定性好,验证了该组合方法在集装箱吞吐量预测中的有效性。

基于组合结构的逻辑回归点击预测算法

随着互联网和广告平台的飞速发展,面对海量的广告信息,为了提升用户点击率,提出一种改进的基于组合结构的逻辑回归点击预测算法LRCS(Logical Regression of Combination Structure)。该算法基于不同类别特征广告受众可能不同的特点,首先,采用FM进行特征组合,产生两类组合特征;其次,将一类特征组合作为聚类算法的输入进行聚类;最后,将另一类特征组合输入由聚类产生的分段GBDT+逻辑回归组合的模型中进行预测。在两个公开数据集中进行了多角度验证,结果表明与其他几类常用的点击预测算法相比,LRCS在点击预测上有一定的性能提升。

土壤Cr含量高光谱反演模型组合优化研究

土壤重金属污染高光谱反演的特征波段提取方法和反演模型的选择是影响反演精度的关键;二者如何优化组合,提高反演精度是目前亟需解决的难题。在华南典型铬(Cr)污染区,采集了92组土壤样品,使用电感耦合等离子体质谱(inductively coupled plasma mass spectrometry,ICP-MS)检测Cr含量,并使用ASD Field Spec4地物光谱仪在实验室收集其高光谱信息。光谱信息预处理采用平滑滤波(SG)+标准正态化(SNV)+二阶微分(SD)变换组合,减弱土壤散射和噪声的影响。选择竞争性自适应重加权采样(CARS)、逐步投影算法(SPA)、无信息变量消除(UVE)、遗传算法(GA)四种算法提取特征波段。选择多元线性回归(MLR)、偏最小二乘法(PLSR)、支持向量回归(SVR)和人工神经网络(ANN)四种反演模型建立特征波段与Cr含量之间的关系。通过对比不同特征波段提取方法和反演模型组合对土壤Cr含量反演的结果发现:采用CARS和UVE特征波段提取方法可以显著提高PLSR、MLR和SVR模型的预测效果;SPA方法能够提高ANN模型的预测效果;通过SG+SNV+SD+CARS+PLSR组合方式,提取位于800~1000、1400~1700以及2100~2450 nm之间的98个特征波段,建模后模型验证,决定系数R2为0.97,均方根误差RMSE为5.25 mg·kg^(-1),平均绝对误差MAE为4.35 mg·kg^(-1),相对分析误差RPD为3.94,表明该模型在预测土壤Cr含量具有优异的性能。以土壤Cr污染高光谱反演为例,通过比较不同特征波段提取方法与反演模型组合的反演精度,确定最优模型,为小样本土壤重金属污染反演的建模提供了思路。