An Electrocardiogram Signal Classification Algorithm Based on Improved Deep Residual Shrinkage Networks北大核心CSCD

心电信号分类是医疗保健领域的重要研究内容.针对大多数方法不能很好地降低样本数量少的类别漏诊率,以及降低预处理操作的复杂性问题,提出了一种基于改进深度残差收缩网络(IDRSN)的心电信号分类算法(即DRSL算法).首先,使用合成少数类过采样技术(SMOTE)扩充数量少的类别样本,从而解决了类不平衡问题;其次,利用改进深度残差收缩网络提取空间特征,其残差模块可以避免网络层加深造成的过拟合,压缩激励和软阈值化子网络可以提取重要局部特征并自动去除噪声;然后,通过长短期记忆网络(LSTM)提取时间特征;最后,利用全连接网络输出分类结果.在MIT-BIH心律失常数据集上的实验结果表明,该算法的分类性能优于IDRSN、DRSN、GAN+2DCNN、CNN+LSTM_ATTENTION、SE-CNN-LSTM分类算法.     

基于支持向量机混合采样的不平衡数据分类方法

利用传统支持向量机(SVM)对不平衡数据进行分类时,由于真实的少数类支持向量样本过少且难以被识别,造成了分类时效果不是很理想.针对这一问题,提出了一种基于支持向量机混合采样的不平衡数据分类方法(BSMS).该方法首先对经过支持向量机分类的原始不平衡数据按照所处位置的不同划分为支持向量区(SV),多数类非支持向量区(MNSV)以及少数类非支持向量区(FNSV)三个区域,并对MNSV区和FNSV区的样本做去噪处理;然后对SV区分类错误和部分分类正确且靠近决策边界的少数类样本重复进行过采样处理,直到找到测试结果最优的训练数据集;最后有选择的随机删除MNSV区的部分样本.实验结果表明:方法优于其他采样方法.     

稀疏交叉熵粗糙集DDRBM-DNNS高校科研能力评估

针对高校科研水平深度学习网络训练评价中存在评价特征同质化现象,造成评估结果精度不高的问题,提出稀疏交叉熵粗糙集双向受限制深度玻尔兹曼机(DDRBM-DNNS)高校科研能力评估方法.首先,考虑采用受限制玻尔兹曼机(RBM)和稀疏交叉熵惩罚参数对深度学习网络进行改进,实现深度学习网络特征训练同质化现象的削弱;同时,针对输入数据的预处理问题,考虑基于粗糙集的前置预处理方式实现,在维持数据输入信息完整前提下,实现输入样本数据的有效归约,进而实现样本处理量的简化,有利于深度学习网络收敛过程的提速;最后,利用所提算法对高校科研水平进行评价,实验数据显示,所提评价模型具备更高的评估精度和更快运算效率.     

面向不平衡数据集的一种基于聚类的欠采样方法

针对不平衡数据集分类问题,提出了一种基于聚类的欠采样方法.分别取不同的聚类个数,对训练集中的多数类样本进行若干次聚类,然后用聚类中心作为多数类样本,与少数类样本构成若干个新的训练集,之后用这些训练集训练分类器,剔除具有错误分类倾向的分类器,最后对分类结果进行投票.仿真实验对几种欠采样方法进行比较.实验采用16个平衡率不一的数据集进行测试.理论分析与实验结果表明:提出的基于聚类的欠采样方法能有效地改善不平衡数据集的不平衡性.     

采样控制系统的稳定性分析新方法

针对采样控制系统的渐近稳定性问题.结合整个采样间隔[t_k,t_(k+1))的特征信息和N阶B-L (Bessel-Legendre)不等式,首次提出一个N-相关的双边闭环L-K (LyapunovKrasovskii)泛函.利用N阶B-L不等式估计L-K泛函的导数,建立了采样控制系统的渐近稳定性新条件.最后通过2个数值算例验证了所得条件的有效性和优越性.     

超高维数据特征筛选方法综述

随着数据收集和存储能力的大幅提高,超高维数据~([9]),即数据维数伴随着样本呈指数增长,频繁出现在许多科学邻域.此时,惩罚类变量选择方法普遍遭遇三个方面的挑战:计算的复杂性,统计的准确性以及算法的稳定性.Fan和Lv~([9])首先提出超高维特征筛选的方法,并在近十多年取得大量研究成果,成为当今统计最热点的研究邻域.本文主要从带模型假设,包含参数、非参数半参数模型假定的筛选方法、无模型假设的筛选以及特殊数据的筛选方法四个角度来介绍超高维筛选相关工作,并简要探讨目前超高维筛选方法存在的问题以及未来可能的研究方向.     

基于条件分布的超高维特征筛选

特征筛选方法是处理超高维数据的一种快速有效的降维方法.针对超高维判别分类数据,提出一种改进的超高维特征筛选方法,方法不需要特定的模型假定;可以处理多分类响应变量情形;可适用于离散型或连续型协变量情形;对服从重尾分布的协变量,方法仍具有较好的稳健性.从理论上证明了所提出特征筛选方法满足确定筛选性和指标排序相合性,并通过数值模拟和实例分析在有限样本条件下验证了方法的有效性.     

基于贝叶斯统计方法的两总体基因表达数据分类

在疾病的诊断过程中,对疾病的精确分类是提高诊断准确率和疾病治愈率至 关重要的一个环节,DNA芯片技术的出现使得我们从微观的层次获得与疾病分类及诊断 密切相关的基因功能信息．但是DNA芯片技术得到的基因的表达模式数据具有多变量小 样本特点,使得分类过程极不稳定,因此我们首先筛选出表达模式发生显著性变化的基因 作为特征基因集合以减少变量个数,然后再根据此特征基因集合建立分类器对样本进行分 类．本文运用似然比检验筛选出特征基因,然后基于贝叶斯方法建立了统计分类模型,并 应用马尔科夫链蒙特卡罗(MCMC)抽样方法计算样本归类后验概率．最后我们将此模型 应用到两组真实的DNA芯片数据上,并将样本成功分类．     

利用少量样本集成吸毒人员预测模型的方法

为了从"潜在"吸毒人员中准确的锁定嫌疑个体,首先甄选样本并利用改进的特征缩放方法得到组合升维的基本样本集,然后对正样本进行过采样、负样本进行样本分片,修正了正、负样本集的不均衡性,最后利用神经网络集成预测吸毒人员的模型,由正负样本集进行训练和验证,提出了基于查准率、准确率、召回率的"三率判断法"的评估模型并进行模拟验证,给出了相关政策建议.     

我国食品卫生安全风险评估模型

食品卫生安全关系到国计民生,建立适合于我国的食品卫生安全保障体系尤为重要.人群食物摄入量模型用"对数正态分布"估计食物摄入量分布,在对国家依据气候、地理、经济等因素进行地域划分的基础上提出了能够保证小样本空间条件下采样均匀的"层次化样本分配模型"用于采样点的选取和各采样点样本数量的分配;污染物分布模型用"贝塔分布"估计食物中的污染物分布,并采用样本在时间轴上平移的方法解决了数据在时间上的不完整性问题;风险评估模型基于人群食物摄入量和食物污染物分布模型,提出"矩元法"对污染物摄入量进行评估,并给出了模型数值化求解和右分位点确定的相关算法.实验表明,风险评估模型所确定的右分位点能有效地反映当前食品卫生安全状况.     

Estimating freeway traffic measures from mobile phone location data

The worldwide propagation of mobile phone and the rapid development of location technologies have provided the chance to monitor freeway traffic conditions without requiring extra infrastructure investment. Over the past decade, a number of research studies and operational tests have attempted to investigate the methods to estimate traffic measures using information from mobile phone. However, most of these works ignored the fact that each vehicle has more than one phone due to the rapid popularity of mobile phone. This paper considered the circumstance of multi-phones and proposed a relatively simplistic clustering technique to identify whether phones travel in the same vehicle. By using this technique, mobile phone data can be used to determine not only speed, but also vehicle counts by type, and therefore density. A complex simulation covering different traffic condition and location accuracy of mobile phone has been developed to evaluate the proposed approach. Simulation results indicate that location accuracy of mobile phone is a crucial factor to estimate accurate traffic measures in case of a given location frequency and the number of continuous location data. In addition, traffic demand and clustering method have a certain effect on the accuracy of traffic measures.     

香港银行间同业拆借利率的非参数统计分析

首次利用短期利率模型,分析香港银行同业拆借利率(Hibor),揭示了最近十年内香港银行同业拆借利率的基本特征.初步分析表明,Hibor数据的平稳性不能保证,因此采用了非参数统计方法.利用bandi文章中的方法,给出了函数的漂移项和扩散项的非参数估计,同时还得到了过程的局部时估计.通过实证分析,发现香港银行间同业拆借利率在2006至2015年间,以2009年为界,前后两个时间段的数据表现出不同的特征,样本数据的局部时函数也表现为双峰分布.     

影响用户换手机的关键因素及其效应研究——基于中国移动西安分公司的样本数据

影响用户换手机的因素有客观因素和主观因素,本研究基于中国移动西安分公司的用户信息数据,根据用户的实际消费和相关非线性多元回归理论,探索了影响用户换手机的客观因素及其效应。研究发现:较长的总通话次数、较大的平均流量和较小的手机屏幕(在用手机)等因素能促进用户换手机;换手机的主要驱动力不是经济能力的差异而是在用手机的屏幕大小;用户的年龄差异对是否换手机的影响可以忽略;城乡标识对是否换手机的影响显著,用户品牌的影响不显著。另外,手机捆绑销售方式的存在,本研究最后站在制造商和运营商的角度,根据以上影响效果给出了提高手机制造商和运营商竞争力的相关建议。     

Telecom service provider portal: Revenue sharing and outsourcing

Early mobile phones only provided voice transmission, for a fee. They have now evolved into voice and online data portals for providing additional services through 3rd party vendors. These service providers (vendors) are given access to a customer base “owned” by the mobile phone companies, for a fee. Typically customers make two payments: to the mobile phone company for phone services and to the 3rd party vendors for specific services bought from them. Variations to the above business model may involve outsourcing the online portal and/or acquiring customers from other independent portals. For these scenarios, we study how the fees for phone service and customer access are established and how they may relate to the prices of vendor services, and which services should be located on the portal - all in a game-theoretic context. Our results prove that it is possible to reorganize revenue flows through an invoicing process that may benefit the mobile network operator more than the other parties. In addition, we establish optimality in terms of the number of vendors on the portal, and determine a rank-ordering of vendors for their inclusion into the portal.     

Linear feature-weighted support vector machine

The existing support vector machines (SVMs) are all assumed that all the features of training samples have equal contributions to construct the optimal separating hyperplane. However, for a certain real-world data set, some features of it may possess more relevances to the classification information, while others may have less relevances. In this paper, the linear feature-weighted support vector machine (LFWSVM) is proposed to deal with the problem. Two phases are employed to construct the proposed model. First, the mutual information (MI) based approach is used to assign appropriate weights for each feature of the whole given data set. Second, the proposed model is trained by the samples with their features weighted by the obtained feature weight vector. Meanwhile, the feature weights are embedded in the quadratic programming through detailed theoretical deduction to obtain the dual solution to the original optimization problem. Although the calculation of feature weights may add an extra computational cost, the proposed model generally exhibits better generalization performance over the traditional support vector machine (SVM) with linear kernel function. Experimental results upon one synthetic data set and several benchmark data sets confirm the benefits in using the proposed method. Moreover, it is also shown in experiments that the proposed MI based approach to determining feature weights is superior to the other two mostly used methods.     

基于用户选择的产品属性权重确定及对TOPSIS法的改进

传统TOPSIS评价法具有权重主观确定、更新方案产生逆序等不足,本文尝试基于用户参与选择的客观行为数据和产品属性数据以改进属性权重的确定;通过极差变换以改进初始矩阵的标准化及最优点、最差点的确定,以消除TOPSIS法评价结果可能出现的逆序问题。最后,基于改进后的TOPSIS法应用于不同品牌手机综合评价的实例验证,结果显示改进后的TOPSIS评价法更具可行性及有效性。     

Modeling of degradation data via wiener stochastic process based on acceleration factor constant principle

This paper proposes a systematic method of modeling accelerated degradation data based on the acceleration factor constant principle. Wiener stochastic process is considered because it is the most extensively used for degradation modeling. For the Wiener stochastic processes with three different time functions, the parameter relationships, which should be satisfied under any two different stress levels, are deduced according to the acceleration factor constant principle. The deduced parameter relationships indicate the stress-related parameters, which are applied to establish accurate accelerated degradation models. In addition, the deduced parameter relationships provide a guidance to test the consistency of the degradation mechanisms under different stress levels. A hypothesis method based on Analysis of Variance is adopted to identify the accelerated stress levels with different degradation mechanism. The degradation data under these stress levels should not be used to assess the product's reliability. The methods of validating accelerated degradation models and reliability assessments are also proposed. The simulation results prove the feasibility and effectiveness of the proposed methods. From the numerical example, it is concluded that the accelerated degradation model established based on the acceleration factor constant principle is more credible and accurate.     

交互式点云建模系统

设计和实现了一个用于车载激光扫描点云处理的交互式建模系统.其原理是:在真实建筑物激光扫描数据的基础上,通过RANSAC实现自动化点云几何特征提取预处理,采用交互引导的方式辅助特征提取,采用随机算法计算边界特征,并通过构建多边形边界邻接图,计算建筑物几何模型.除特征提取外,涉及到的计算都是实时的.与传统处理方法相比,该系统具有速度快、精度高等优点.     

基于核主元提取的支持向量机辨识

将一种基于特征提取的ε-不灵敏支持向量机方法用于非线性系统辨识.对输入输出数据首先进行核主元特征提取,将特征提取后的数据作为支持向量机的训练数据.将该方法与基于主元特征提取的方法和直接应用ε-不灵敏支持向量机的方法进行含噪和不含噪情况下的仿真比较,结果表明,方法的拟合性能和抗干扰能力优于其他两种方法.     

Motorway traffic parameter estimation from mobile phone counts

In this paper a new method for real time estimation of vehicular flows and densities on motorways is proposed. This method is based on fusing traffic counts with mobile phone counts. The procedure used for the estimation of traffic flow parameters is based on the hypothesis that “instrumented” vehicles can be counted on specific motorway sections and traffic flow can be measured on entrance and exit ramps. The motorway is subdivided into cells, assuming that mobile phones entering and exiting every cell can be counted during the observation period. An estimate of “instrumented” vehicle concentration is obtained and propagated on the network in time and space. This allows one to estimate traffic flow parameters by sampling “instrumented” traffic flow parameters using a “concentration” (the ratio of the densities of instrumented vehicles to the density of overall traffic) propagation mechanism.