影响ETM影像土地利用/覆盖分类精度因素的研究

训练样本量、辅助数据和分类法是影响土地利用/覆盖分类精度的3个主要因素,通过找到这3个因素的最佳组合方式以提高分类精度,分别在25%、50%、75%、100%样本量下,加入NDVI、DEM和纹理均值特征作为辅助数据,比较了分类回归树、支持向量机、最大似然法3种分类法的效果,探讨了训练样本、辅助数据以及分类技术对土地利用/覆盖分类精度的影响。结果表明：支持向量机总体分类精度较高,在相同样本量和没有有效辅助数据的情况下,SVM可以获得最佳的分类结果,总体分类精度在85%以上;在进行分类时,加入NDVI和纹理均值特征使分类回归树分类精度提高了2.82%,说明该方法对有效辅助数据的加入较为敏感;在获取的训练样本集有限而可获取有效的辅助数据时,应优先考虑利用分类回归树进行土地利用/覆盖分类。     

基于CART算法的带钢抗拉强度影响因素研究

CART是数据挖掘的一种全新的优越的分类工具,是一种产生二叉决策树的技术。由CART模型构建的预测树在很多情况下比常用的统计方法构建的代数学预测准则更加准确,且数据越复杂、变量越多,算法的优越性就越显著。基于信息论的CART算法已在国际上被广泛的应用。阐述了CART(Classification and Regression Tree)算法的基本原理和主要特征,并介绍了连退的基本工艺流程和抗拉强度的概念。通过理论分析,选取了某钢厂连退机组生产过程中影响带钢抗拉强度的重要因素,视为模型的决策属性。根据CART算法,挖掘出三大影响抗拉强度的重要属性,建立了抗拉强度的评估规则。研究表明,CART算法有效的处理了因子之间的非线性关系,建立了可靠、有效的规则,为带钢生产提供了良好的决策依据。     

分类回归树在高校计算机联考数据分析中的应用

分析了分类回归树的基本结构及树的构建方法,并将其应用到高校计算机联考的数据分析中。结果表明,该方法能够较好的对数据进行分类,所生成的分类规则有助于今后教学和学生工作的开展。     

An alternative objective function for fitting regression trees to functional response variables

Analyses of systems that can be represented by functional responses are becoming common in many scientific disciplines. Functional regression trees (FRT) provide a methodology for modelling such systems. Recent work has focused on fitting models where the response variable is a probability density function, using a splitting criterion that is based on the sum of dissimilarities between the densities. We suggest a different criterion based on deviations of the densities from their mean. We provide motivation and justification for this criterion, and demonstrate its superior performance using an extensive simulation exercise. We discuss the computational aspects of the FRT procedure and show that substantial speed gains can be made through use of a dissimilarity matrix. Our results show that the proposed splitting criterion outperforms both the original and a splitting criterion based on Euclidean distance. Pointwise standard error curves for a predicted functional response can be generated through the fitting procedure, which we demonstrate in a case study with a forestry data set. Supplementary materials are available.     

基于互补模型的汉语韵律间断自动检测

自动韵律间断检测和标注对语音理解和语音合成有十分重要的作用。提出了利用声学、词典和语法相关特征的互补模型方法检测汉语韵律间断。该方法具有下列优点:(1)摒弃了声学相关特征和词典、语法相关特征的独立性假设;(2)互补模型方法不仅在特征层上利用当前音节的上下文信息,而且在模型层次上利用了当前音节的上下文信息。在ASCCD语料库上验证了该方法能够获得90.34%的韵律间断的检测准确率,较基线系统有 6.09%的提高。     

Mass appraisal of residential apartments: An application of Random forest for valuation and a CART-based approach for model diagnostics

To the best knowledge of authors, the use of Random forest as a potential technique for residential estate mass appraisal has been attempted for the first time. In the empirical study using data on residential apartments the method performed better than such techniques as CHAID, CART, KNN, multiple regression analysis, Artificial Neural Networks (MLP and RBF) and Boosted Trees. An approach for automatic detection of segments where a model significantly underperforms and for detecting segments with systematically under- or overestimated prediction is introduced. This segmentational approach is applicable to various expert systems including, but not limited to, those used for the mass appraisal.     

决策树分类技术研究

决策树分类是一种重要的数据分类技术。ID3、C4．和EC4．5是建立决策树的常用算法，但目前国内对一些新的决策树分类算法研究较少。为此，在消化大量文献资料的基础上，研究了CART、SLIQ、SPRINT、PUBLIC等新算法，对各种决策树分类算法的基本思想进行阐述，并分析比较了各种算法的主要特性，为数据分类研究者提供借鉴。     

Vehicle classification with single multi-functional magnetic sensor and optimal MNS-based CART

To take advantages of magnetic sensor technology in terms of cost, size, weight, power consumption and wireless communication, a wireless multi-functional magnetic sensor was designed and developed. Then, a novel method with single multi-functional magnetic sensor and optimal Minimum Number of Split-sample (MNS)-based Classification and Regression Tree (CART) algorithm was proposed in this paper to classify on-road vehicles. The sensor was deployed on the road to acquire real-time vehicle waveform data. The decision tree model based on CART algorithm was used to execute on-line vehicle classification in the sensor node. Eight speed-independent time-domain waveform features were extracted as the model inputs. This paper trained the decision tree model by using vehicle samples derived from the multi-functional magnetic sensor and pruned the optimal decision tree with a Minimum Error Pruning (MEP) rule to obtain an optimal pruning tree which is more robust to new samples. Some experiments were implemented by different sample sets and classification methods. The results showed that the proposed method achieved on-line vehicle classification in the sensor node. For the field sample sets with two vehicle classes, the average accuracy rates of test samples were 88.9% and 94.4% in the original samples and swapping samples respectively. Besides higher accuracy, the method also has a better sample robustness, which is easy to classify new samples. The comparison results of current methods also showed that the proposed method has some advantages in aspects of accuracy rate, sample robustness and execution time.     

Assessment of resistance spot welding quality based on ultrasonic testing and tree-based techniques

Classification and regression tree (CART) and random forest techniques were proposed as pattern recognition tools for classification of ultrasonic oscillograms of resistance spot welding (RSW) joints. The results showed that CART models produced an acceptable error rate with high interpretability. These features may be used to understand and control the decision processes, instruct other human operators, compare margins of safety or modify them depending on the criticality of the industrial process. Compared with CART trees, random forests reduced the error rate at the cost of decreasing decision interpretability. The use of the agreement of the forest was proposed as a measure to reduce the workload of human operators, who would only have to focus on the analysis of ultrasonic oscillograms that are difficult to interpret.     

基于氧气光谱吸收的被动测距中无云天空背景辐射特性研究

氧气吸收率是利用氧气A 吸收带进行被动测距技术计算的核心,将包含氧气A 吸收带在内的无云天空背景辐射和黑体辐射作为研究对象,利用CART 软件模拟计算了不同观测天顶角、不同时段、不同太阳天顶角的无云天空背景氧气吸收率分布,并与不同观测天顶角、不同距离下的黑体辐射氧气吸收率进行了比较分析,结果表明:当探测距离大于3 km 时,黑体辐射的氧气吸收率大于无云天空背景辐射氧气吸收率,所以根据不同观测条件设置被动测距的氧气吸收率阈值,可提高对目标的探测概率,降低背景辐射对被动测距的影响。