基于模糊分割和邻近对的支持向量机分类器

支持向量机算法对噪声点和异常点是敏感的,为了解决这个问题,人们提出了模糊支持向量机,但其中的模糊隶属度函数需要人为设置。提出基于模糊分割和邻近对的支持向量机分类器。在该算法中,首先根据聚类有效性用模糊c-均值聚类算法分别对训练集中的正负类数据聚类;然后,根据聚类结果构造c个二分类问题,求解得c个二分类器;最后,用邻近对策略对样本点进行识别。用4个著名的数据集进行了数值实验,结果表明该算法能有效提高带噪声点和异常点数据集分类的预测精度。     

Fuzzy c-means based support vector machines classifier for perfume recognition

Identification of more than three perfumes is very difficult for the human nose. It is also a problem to recognize patterns of perfume odor with an electronic nose that has multiple sensors. For this reason, a new hybrid classifier has been presented to identify type of perfume from a closely similar data set of 20 different odors of perfumes. The structure of this hybrid technique is the combination of unsupervised fuzzy clustering c-mean (FCM) and supervised support vector machine (SVM). On the other hand this proposed soft computing technique was compared with the other well-known learning algorithms. The results show that the proposed hybrid algorithm’s accuracy is 97.5% better than the others.     

Mahalanobis distance based on fuzzy clustering algorithm for image segmentation

Conventional Fuzzy C-means (FCM) algorithm uses Euclidean distance to describe the dissimilarity between data and cluster prototypes. Since the Euclidean distance based dissimilarity measure only characterizes the mean information of a cluster, it is sensitive to noise and cluster divergence. In this paper, we propose a novel fuzzy clustering algorithm for image segmentation, in which the Mahalanobis distance is utilized to define the dissimilarity measure. We add a new regularization term to the objective function of the proposed algorithm, reflecting the covariance of the cluster. We experimentally demonstrate the effectiveness of the proposed algorithm on a generated 2D dataset and a subset of Berkeley benchmark images.     

基于自适应马氏距离的模糊c均值算法

经典的模糊c均值（FCM）算法是基于欧氏距离的,它只适用于球型结构的聚类,且在处理高维的数据集时,分错率增加。针对以上两个问题,提出了一种新的聚类算法（FCM-M）,它将马氏距离与模糊c均值相结合,并在目标函数中引进一个协方差矩阵的调节因子,利用马氏距离的优点,有效地解决了FCM算法中的缺陷,并利用特征值、特征矢量及伪逆运算来解决马氏距离中遇到的奇异问题。通过数据聚类和图像分割两组实验,证实了该方法的可行性和有效性。     

基于构造映射的支持向量分类机

构造了基于马氏距离和Cupula函数的距离映射和条件概率映射,将高维空间中的样本转化为二维空间中的新样本,并对新样本进行简易处理,构建了可分支持向量分类机,其特点是简单,易于实现。实验表明取得了较好的分类效果。     

Improved spatial fuzzy c-means clustering for image segmentation using PSO initialization,Mahalanobis distance and post-segmentation correction

In this paper, we propose an improvement method for image segmentation using the fuzzy c-means clustering algorithm (FCM). This algorithm is widely experimented in the field of image segmentation with very successful results. In this work, we suggest further improving these results by acting at three different levels. The first is related to the fuzzy c-means algorithm itself by improving the initialization step using a metaheuristic optimization. The second level is concerned with the integration of the spatial gray-level information of the image in the clustering segmentation process and the use of Mahalanobis distance to reduce the influence of the geometrical shape of the different classes. The final level corresponds to refining the segmentation results by correcting the errors of clustering by reallocating the potentially misclassified pixels. The proposed method, named improved spatial fuzzy c-means IFCMS, was evaluated on several test images including both synthetic images and simulated brain MRI images from the McConnell Brain Imaging Center (BrainWeb) database. This method is compared to the most used FCM-based algorithms of the literature. The results demonstrate the efficiency of the ideas presented.     

Various approaches to speed up Mahalanobis distance classifier

In this paper, we propose various approaches to reduce the CPU time requirement of Mahalanobis distance classifier (MDC). Based on the results discussed in this paper, we have found that the Ensemble average approach and Quadratic form range theorem are superior to other algorithms. The algorithms based on these two approaches are found to give a maximum speed of 7 and 9 with MSS and TM data, respectively.     

基于模糊支持向量机与决策树的文本分类器

针对模糊支持向量机在文本分类应用中的隶属度函数确定问题,提出了一种基于模糊支持向量机与决策树的文本分类器的构建方法。该方法不仅考虑了样本与类中心之间的关系,还根据传统支持向量机中包含支持向量且平行于分类面的平面构建切球,来确定类中各个样本之间的关系,由样本点与球的位置关系计算其隶属度,可以合理地区分有效样本和噪音、孤立点样本。并与决策树方法相结合,实现多类分类。实验结果表明,该方法具有良好的分类效果。     

基于模糊sigmoid核的支持向量机回归建模

支持向量机中对核函数的要求为对称的半正定矩阵.来自于神经网络的sigm o id核函数在其参数满足一定条件时才成为半正定矩阵,但是这种核函数在SVM中却有很多成功的应用.本文将sigm o id核函数与模糊逻辑相结合并使其模糊化,从而简化了SVM的计算并便于用硬件实现.通过对混沌时间序列预测以及图像去噪滤波器两个实例的实验研究发现,使用模糊sigm o id核函数可以使SVM回归建模在损失较小精度的代价下,较大地降低平均CPU执行时间,便于硬件实现     

Fuzzy fault diagnosis based on fuzzy robust v-support vector classifier and modified genetic algorithm

This paper presents a new version of fuzzy support vector classifier machine (SVM) which can penalize those hybrid noises to forecast fuzzy nonlinear system. Since there exist some problems of uncertain data in many actual forecasting problem, the input variables are described as fuzzy numbers by fuzzy comprehensive evaluation. To solve the shortage of ε-insensitive loss function for hybrid noises such as singularity points, biggish magnitude noises and Gaussian noises, a novel robust loss function is proposed in this paper. Then by the integration of the triangular fuzzy theory, v-SVC and robust loss function theory, fuzzy robust v-SVC (FRv-SVM) which can penalize those hybrid noises is proposed. To seek the optimal parameters of FRv-SVC, genetic algorithm is also proposed to optimize the unknown parameters of FRv-SVC. The results of the application in fuzzy car assembly line system diagnosis confirm the feasibility and the validity of the FRv-SVC model. Compared with other SVC methods, FRv-SVC method has better classifier precison for small sample with hybrid noises.     

基于自适应距离度量的最小距离分类器集成

提出了一种基于自适应距离度量的最小距离分类器集成方法,给出了个体分类器的生成方法。首先用Bootstrap技术对训练样本集进行可重复采样,生成若干个子样本集,应用生成的子样本集建立自适应距离度量模型,根据建立的模型对子样本集进行训练,生成个体分类器。在集成中,将结果用相对多数投票法集成最终的结论。采用UCI标准数据集实验,将该方法与已有方法进行了性能比较,结果表明基于自适应距离度量的最小距离分类器集成是最有效的。     

基于马氏距离和灰色分析的缺失值填充算法

针对kNN算法中欧氏距离具有密度相关性敏感的缺点,提出综合马氏距离和灰色分析方法代替kNN算法中欧式距离的新算法,应用到缺失数据填充方面。其中马氏距离能解决密度相关明显的数据集,灰色分析方法能处理密度相关不明显的情况。因此,该算法能很好处理任何数据集,实验结果显示,算法在填充结果上明显优于现有的其他算法。     

基于减法聚类改进的模糊c-均值算法的模糊聚类研究

针对模糊c-均值(FCM)聚类算法受初始聚类中心影响,易陷入局部最优,以及算法对孤立点数据敏感的问题,提出了解决方案:采用快速减法聚类算法初始化聚类中心,为每个样本点赋予一个定量的权值,用来区分不同的样本点对最终的聚类结果的不同作用,为提高聚类速度采用修正隶属度矩阵的方法,并将算法与传统的FCM相比.实验结果表明,该算法较好地解决了初值问题,与随机初始化方法相比,迭代次数少、收敛速度快、具有较好的聚类结果.     

基于加权参数马氏距离的硬件木马检测

针对常规马氏距离判别方法对硬件木马检测率低下或失效的情况,通过对芯片运行时产生的功耗特征进行建模分析,提出了一种基于加权与双参数变换的优化判别方案.首先对标准与待测样本矩阵进行参数调整并在判别公式中加入样本矩阵特征向量权重,运用MATLAB实现参数的最优组合,最后在置信度99％的条件下计算待测样本的硬件木马检测率.对待测FPGA植入占比0.3％的硬件木马,并通过FPGA硬件木马检测平台验证表明,在常规马氏距离判别方法检测率只有43.56％的情况下,该方法检测率达到了85.14％.     

基于模糊核c-means 算法的位置指纹聚类

提出一种针对位置指纹的模糊核c-means聚类算法.将位置指纹归结为一种服从正态分布的区间值数据以反映接入点信号强度采样值的不确定性,通过区间中值和大小确定的正态分布函数将位置指纹映射为特征空间中的一点,并在该特征空间中采用基于核方法的模糊c-means算法对其进行聚类.通过ZigBee定位实验表明,该方法对于位置指纹的分类效果明显好于基于信号强度平均值的c-means聚类,可在保证定位精度的前提下有效降低定位的计算量.     

Incipient sensor fault isolation based on augmented Mahalanobis distance

Incipient sensor fault diagnosis is important to an efficient and optimal operating condition for modern industrial systems. Recently, a new fault detection index called augmented Mahalanobis distance (AMD) has been proposed in our previous work for incipient fault detection. Following detection, fault isolation is also quite desired so as to investigate root causes of the occurred fault. In the present work, the AMD statistic is first revisited and a geometric illustration of AMD is provided, which intuitively shows its superiority for incipient fault detection. Then, with available fault direction information, an incipient sensor fault isolation approach is proposed. Its fault isolability condition is analyzed theoretically and compared with that of the conventional method. For complex sensor faults whose fault direction information is unknown, a corresponding fault isolation strategy is also briefly discussed. Case studies on a high-speed train air brake system and the continuous stirred tank reactor (CSTR) process are carried out, which demonstrate the effectiveness of the AMD based fault detection and isolation methods, in comparison with conventional approaches.     

一类直接构造的模糊多类支持向量分类器

针对直接多类分类方法,提出了一种新的基于直接构造多类SVM分类器的模糊多类支持向量机算法FCS-SVM。在算法中,重构了优化问题及其约束条件,以及Lagrange公式,并进行了推导。通过在标准数据集上的几个实验,对这些算法进行了比较分析。实验结果表明提出的算法可以得到比较理想的分类精度。     

基于邻域信息的改进模糊c均值脑MRI分割

在脑图像分割中,噪声或异常值的干扰往往会使得图像的质量下降。而传统的模糊c均值算法存在一定的缺限,容易受初始值的影响,这给医生准确识别和提取脑组织带来很大的麻烦。针对这些问题,提出一种基于用马尔可夫模型构建的图像像素点邻域的改进模糊c均值图像分割方法。首先,用遗传算法（GA）确定初始的聚类中心;然后,改变目标函数的表达方式,通过在目标函数中添加修正项来改变隶属度矩阵的计算方式,并用约束系数对其来调节;最后,由马尔可夫随机域来表达邻域像素的标号信息,并利用马尔可夫随机场（MRF）的最大化条件概率来表示像素的邻域,增强了抗噪性。实验结果显示,该方法拥有较好的抗噪性,可以降低误分割率,在对脑图像分割时具备较高的分割精度。分割后的图像平均精度可达：JS（Jaccard Similarity）指标为82.76%,Dice指标为90.45%,Sensitivity指标为90.19%;同时,对脑图像边界处的分割更加清晰,分割后的图像更加接近于标准分割图像。     

An integrated approach to fuzzy learning vector quantization andfuzzy c-means clustering

Derives an interpretation for a family of competitive learning algorithms and investigates their relationship to fuzzy c-means and fuzzy learning vector quantization. These algorithms map a set of feature vectors into a set of prototypes associated with a competitive network that performs unsupervised learning. Derivation of the new algorithms is accomplished by minimizing an average generalized distance between the feature vectors and prototypes using gradient descent. A close relationship between the resulting algorithms and fuzzy c-means is revealed by investigating the functionals involved. It is also shown that the fuzzy c-means and fuzzy learning vector quantization algorithms are related to the proposed algorithms if the learning rate at each iteration is selected to satisfy a certain condition     

基于聚类和马氏距离的多角度SURF图像匹配算法

对具有不同旋转角度和变化的图像进行匹配是图像识别中的技术难点,SURF算法在多角度图像的特征点检测和匹配过程中存在易受噪声点干扰、产生误匹配从而导致匹配效率低等不足。结合聚类和马氏距离,提出一种改进的多角度SURF图像匹配算法。首先利用聚类算法对原有算法提取的特征点进行噪声剔除处理,生成新的特征点数据集;然后利用马氏距离能够有效考虑整体相关性及其具有仿射不变性等特点,将SURF算法中的欧式距离用马氏距离替代。实验应用于多角度图像匹配时,改进算法较原SURF算法在匹配效率和准确率上有明显提高。