DATA PREPROCESSING AND RE KERNEL CLUSTERING FOR LETTER

Many classifiers and methods are proposed to deal with letter recognition problem. Among them, clustering is a widely used method. But only one time for clustering is not adequately. Here, we adopt data preprocessing and a re kernel clustering method to tackle the letter recognition problem. In order to validate effectiveness and efficiency of proposed method, we introduce re kernel clustering into Kernel Nearest Neighbor classification(KNN), Radial Basis Function Neural Network(RBFNN), and Support Vector Machine(SVM). Furthermore, we compare the difference between re kernel clustering and one time kernel clustering which is denoted as kernel clustering for short. Experimental results validate that re kernel clustering forms fewer and more feasible kernels and attain higher classification accuracy.

A Fast and Effective Multiple Kernel Clustering Method on Incomplete Data

Multiple kernel clustering is an unsupervised data analysis method that has been used in various scenarios where data is easy to be collected but hard to be labeled.However,multiple kernel clustering for incomplete data is a critical yet challenging task.Although the existing absent multiple kernel clustering methods have achieved remarkable performance on this task,they may fail when data has a high value-missing rate,and they may easily fall into a local optimum.To address these problems,in this paper,we propose an absent multiple kernel clustering(AMKC)method on incomplete data.The AMKC method rst clusters the initialized incomplete data.Then,it constructs a new multiple-kernel-based data space,referred to as K-space,from multiple sources to learn kernel combination coefcients.Finally,it seamlessly integrates an incomplete-kernel-imputation objective,a multiple-kernel-learning objective,and a kernel-clustering objective in order to achieve absent multiple kernel clustering.The three stages in this process are carried out simultaneously until the convergence condition is met.Experiments on six datasets with various characteristics demonstrate that the kernel imputation and clustering performance of the proposed method is signicantly better than state-of-the-art competitors.Meanwhile,the proposed method gains fast convergence speed.

Multiple Kernel Clustering Based on Self-Weighted Local Kernel Alignment

Multiple kernel clustering based on local kernel alignment has achieved outstanding clustering performance by applying local kernel alignment on each sample.However,we observe that most of existing works usually assume that each local kernel alignment has the equal contribution to clustering performance,while local kernel alignment on different sample actually has different contribution to clustering performance.Therefore this assumption could have a negative effective on clustering performance.To solve this issue,we design a multiple kernel clustering algorithm based on self-weighted local kernel alignment,which can learn a proper weight to clustering performance for each local kernel alignment.Specifically,we introduce a new optimization variable-weight-to denote the contribution of each local kernel alignment to clustering performance,and then,weight,kernel combination coefficients and cluster membership are alternately optimized under kernel alignment frame.In addition,we develop a three-step alternate iterative optimization algorithm to address the resultant optimization problem.Broad experiments on five benchmark data sets have been put into effect to evaluate the clustering performance of the proposed algorithm.The experimental results distinctly demonstrate that the proposed algorithm outperforms the typical multiple kernel clustering algorithms,which illustrates the effectiveness of the proposed algorithm.

A Kernel Clustering Algorithm for Fast Training of Support Vector Machines

A new algorithm named kernel bisecting k-means and sample removal(KBK-SR) is proposed as sampling preprocessing for support vector machine(SVM) training to improve the efficiency.The proposed algorithm tends to quickly produce balanced clusters of similar sizes in the kernel feature space,which makes it efficient and effective for reducing training samples.Theoretical analysis and experimental results on three UCI real data benchmarks both show that,with very short sampling time,the proposed algorithm dramatically accelerates SVM sampling and training while maintaining high test accuracy.

Scaling up Kernel Grower Clustering Method for Large Data Sets via Core-sets

核栽培者是聚类最近 Camastra 和 Verri 建议的方法的一个新奇的核。它证明为各种各样的数据的好性能关于流行聚类的算法有利地设定并且比较。然而，方法的主要缺点是在处理大数据集合的弱可伸缩能力，它极大地限制它的应用程序。在这份报纸，我们用核心集合建议一个可伸缩起来的核栽培者方法，它是比为聚类的大数据的原来的方法显著地快的。同时，它能处理很大的数据集合。象合成数据集合一样的基准数据集合的数字实验显示出建议方法的效率。方法也被用于真实图象分割说明它的性能。

Kernel method-based fuzzy clustering algorithm

The fuzzy C-means clustering algorithm(FCM) to the fuzzy kernel C-means clustering algorithm(FKCM) to effectively perform cluster analysis on the diversiform structures are extended, such as non-hyperspherical data, data with noise, data with mixture of heterogeneous cluster prototypes, asymmetric data, etc. Based on the Mercer kernel, FKCM clustering algorithm is derived from FCM algorithm united with kernel method. The results of experiments with the synthetic and real data show that the FKCM clustering algorithm is universality and can effectively unsupervised analyze datasets with variform structures in contrast to FCM algorithm. It is can be imagined that kernel-based clustering algorithm is one of important research direction of fuzzy clustering analysis.

An Improved Kernel K-Mean Cluster Method and Its Application in Fault Diagnosis of Roller Bearing

For the kernel K-mean cluster method is run in an implicit feature space, the initial and iterative cluster centers cannot be defined explicitly. Against the deficiency of the initial cluster centers selected in the original space discretionarily in the existing methods, this paper proposes a new method for ensuring the clustering center that virtual clustering centers are defined in the feature space by the original classification as the initial cluster centers and the iteration clustering centers are ensured by the further virtual classification. The improved method is used for fault diagnosis of roller bearing that achieves a good cluster and diagnosis result, which demonstrates the effectiveness of the proposed method.

Kernel Generalized Noise Clustering Algorithm

To deal with the nonlinear separable problem, the generalized noise clustering （GNC） algorithm is extended to a kernel generalized noise clustering （KGNC） model. Different from the fuzzy c-means （FCM） model and the GNC model which are based on Euclidean distance, the presented model is based on kernel-induced distance by using kernel method. By kernel method the input data are nonlinearly and implicitly mapped into a high-dimensional feature space, where the nonlinear pattern appears linear and the GNC algorithm is performed. It is unnecessary to calculate in high-dimensional feature space because the kernel function can do it just in input space. The effectiveness of the proposed algorithm is verified by experiments on three data sets. It is concluded that the KGNC algorithm has better clustering accuracy than FCM and GNC in clustering data sets containing noisy data.

Face Recognition Using Fuzzy Clustering and Kernel Least Square

Over the last fifteen years, face recognition has become a popular area of research in image analysis and one of the most successful applications of machine learning and understanding. To enhance the classification rate of the image recognition, several techniques are introduced, modified and combined. The suggested model extracts the features using Fourier-Gabor filter, selects the best features using signal to noise ratio, deletes or modifies anomalous images using fuzzy c-mean clustering, uses kernel least square and optimizes it by using wild dog pack optimization. To compare the suggested method with the previous methods, four datasets are used. The results indicate that the suggested methods without fuzzy clustering and with fuzzy clustering outperform state- of-art methods for all datasets.

Modified possibilistic clustering model based on kernel methods

A novel model of fuzzy clustering using kernel methods is proposed. This model is called kernel modified possibilistic c-means （KMPCM） model. The proposed model is an extension of the modified possibilistic c-means （MPCM） algorithm by using kernel methods. Different from MPCM and fuzzy c-means （FCM） model which are based on Euclidean distance, the proposed model is based on kernel-induced distance. Furthermore, with kernel methods the input data can be mapped implicitly into a high-dimensional feature space where the nonlinear pattern now appears linear. It is unnecessary to do calculation in the high-dimensional feature space because the kernel function can do it. Numerical experiments show that KMPCM outperforms FCM and MPCM.

基于匝道合流数据的自动驾驶汽车安全性测试评价方法

针对自动驾驶汽车测试场景不明确、评价模型主观性强等问题,研究了高速匝道汇入场景下的典型测试场景提取方法和自动驾驶汽车匝道汇入安全性客观评价方法。深入分析了匝道汇入功能场景下的逻辑场景要素,对自然驾驶数据中的自车速度、车间距离、前车车速等逻辑场景要素进行聚类,得到两类典型的匝道汇入测试场景用于自动驾驶汽车的仿真测试。构建多层次自动驾驶汽车评价体系,引入基于自然驾驶数据的核密度估计模型来获取指标最优阈值,建立以最优阈值为参考序列、以层次分析法(AHP)和客观赋权法(CRITIC)为权重输入的灰色关联理评价模型,对自动驾驶汽车在汇入过程中的安全性进行客观评价。评价结果表明:基于核密度估计的灰色关联理论模型评价结果与主观模糊综合分析模型的评价结果相似率达98.01%,验证了客观模型的有效性。

基于核聚类的砂岩图像孔隙分割方法

砂岩孔隙识别是研究孔隙结构的一个重要步骤,采用通用的图像分割算法不易得到理想的图像孔隙分割效果,为此提出了一种使用EfficientNetV2-S模型和核K-Means聚类技术对孔隙进行分割的方法。首先,获得砂岩图像的超像素集合,使用超像素方法预分割输入的致密砂岩图像,构建带标签的孔隙与非孔隙图像库;然后,应用EfficientNetV2-S模型提取砂岩图像的孔隙和非孔隙的语义特征,并结合迁移学习的方法,使用有限的砂岩图像的孔隙和非孔隙样本进行EfficientNetV2-S模型参数学习;最后,设计了一种基于K-Means聚类的区域合并方法——NTK-KCoP方法,根据超像素的语义特征、灰度特征和边缘特征构建目标函数,再由聚类结果合并超像素得到完整的孔隙区域。砂岩CT图像的实验结果验证了所提出的孔隙分割方法的适用性和有效性。

基于形态特征的破碎玉米籽粒识别及检测装置设计

基于K-Means聚类分割算法、颜色空间转换和形态学特征,通过对辽沈地区广泛种植的玉米的面积、面积与周长比、短轴与长轴比、圆形与矩形及五种形态特征的分析,识别出玉米碎粒。同时,设计开发了一种基于同步带轮的玉米籽粒破碎率在线检测装置,用于玉米籽粒破碎率的在线检测与识别。研究结果:①利用K平均算法分割算法将图像分割为目标区域和背景区域;②利用K平均算法分割算法将图像分割为二值化图像,进一步对二值化图像采用形态闭运算填充目标区域的非连通区域,对目标区域进行统计分析,计算出形态特征;③通过多组试验,玉米籽粒破碎的识别率可达94%。

改进DPC聚类算法的离群点检测与解释方法

为解决全局离群点检测方法无法对局部离群点进行检测,以及局部异常因子在面对大量局部离群点时性能下降的问题,利用k近邻(KNN)和核密度估计方法(KDE)提出一种基于改进快速搜索和发现密度峰值聚类算法(KDPC)的离群点检测与解释方法,该方法能够同时对数据点的全局和局部进行分析。首先,利用k近邻和核密度估计方法计算数据点的局部密度,代替传统DPC算法中根据截断距离计算的局部密度。其次,将数据点的k近邻距离之和作为全局异常值,并通过KDPC聚类算法计算簇密度以及数据点的局部异常值。最后,将数据点的全局与局部异常值进行乘积作为最终异常得分,选取异常得分最高的Top-n作为离群点,通过构建全局-局部异常值决策图对全局和局部离群点进行解释。利用人工数据集和UCI数据集进行实验并与10种常用离群点检测方法进行比较。结果表明,该方法对全局和局部离群点都有着较高的检测精度和检测性能,并且AUC方面受k值影响较小。同时,利用该方法对NBA球员数据进行分析讨论,进一步证明了该方法的实用性和有效性。

基于KANN-DBSCAN带宽优化的核密度估计载荷谱外推

针对核密度估计载荷外推全局固定带宽的局限性,提出一种基于KANN-DBSCAN(K-average nearest neighbor density-based spatial clustering of applications with noise)改进带宽取值的核密度估计(kernel density estimation, KDE)载荷外推方法。通过KANN-DBSCAN聚类算法对载荷数据进行分组聚类,采用拇指法求得不同簇间的最优带宽,然后进行核密度估计,再采用蒙特卡洛模拟进行外推。以某电动汽车在用户道路的实测载荷数据为应用对象,对外推方法的合理性进行检验。从统计参数检验量、拟合度检验和伪损伤检验3个指标对外推效果进行评估。结果表明:相比固定带宽的核密度估计外推方法,基于KANN-DBSCSN核密度估计的外推方法获得的外推载荷在统计参数上与实测载荷更为接近,均值、标准差和最大值的误差分别仅为1.9%、 4.3%和1.9%;幅值累计频次曲线拟合度R2均大于0.99,伪损伤均接近1。结果验证了该聚类方法在核密度估计载荷外推的有效性,有助于编制汽车在用户道路上的载荷谱,为具有相似载荷分布特点的机械零部件载荷外推提供了参考。

基于改进FCM的冲压件缺陷图像分割算法

在工业质检过程中,冲压件缺陷图像分割作为缺陷检测的重要环节,直接影响缺陷检测效果。而传统的模糊C均值(FCM)聚类算法未考虑到空间邻域信息,对于噪声干扰较为敏感,导致分割精度较差,且其整体易受初始值的影响,造成收敛速度变慢。针对上述问题,提出一种改进的FCM算法。采用内核诱导距离中的简单两项代替传统的欧氏距离,将原有的空间像素映射到高维特征空间,提高线性可分概率和计算速度;利用图像像素之间的空间相关性,通过引入改进的马尔可夫随机场对FCM目标函数进行修正,提高算法的抗噪能力以及分割精度;采用秃鹰搜索(BES)算法确定FCM的初始聚类中心,提高算法的收敛速度,同时避免算法陷入局部极值的情况。为验证改进FCM算法的性能,选取划分熵、划分系数、Xie_Beni系数以及迭代次数作为评价指标,并与近年来先进的图像分割算法进行对比。实验结果表明,改进FCM算法具有更好的抗噪能力,能得到更好的缺陷分割效果,对工业生产中的冲压件缺陷检测有一定的应用价值。

顾及细分行业的宁波市海洋产业空间演化及动因分析

基于宁波市涉海企业工商登记数据,首先引入机器学习领域的文本分类算法识别各企业在海洋产业中的细分行业类别,进而通过核密度估计和标准差椭圆等分析1992—2022年宁波市海洋产业中各细分行业的空间布局与演化特征,并探究涉海企业选址与全球化、地方化关联过程之间的逻辑关系,最后利用地理探测器分析影响海洋产业布局的因素。结果表明:①研究期内宁波市海洋产业各细分行业核心区数量增加,集聚核心区体系的规模秩序提升且趋于稳定,集聚程度均有提高,行业间的空间关联关系显现,各行业核心区与社会经济实体间存在稳固的匹配关系;②宁波市海洋产业集群化的过程是本地涉海企业参与全球化和地方化的过程,市场和制度环境优化促使要素配置水平提高,地方文化和企业家精神推动行业创新,经济活动嵌入社会网络后加强了行业空间集聚路径的依赖,区位条件的改变和专业化分工成为海洋产业空间布局调整的重要动力,经济全球化对涉海企业空间格局调整的灵活性具有深刻影响;③行政可达性、城镇化水平和交通便利程度是影响产业布局的重要因素,经济联系密切的产业在空间上相互影响,影响因素交互会加大对产业空间格局的解释力。

基于Kernel K-means的负荷曲线聚类

电力负荷曲线聚类是配用电系统的基础,对负荷管理具有重大意义。采用基于核方法的聚类算法提高负荷曲线聚类的准确性,通过点积的方式构造核矩阵,再将数据映射到高维空间中进行聚类,进而加大数据的可分性。同时,针对核矩阵的规模大、计算复杂的问题,提出使用核主成分与缩减矩阵规模对该方法进行优化。实验过程中采用美国能源部开发能源信息网站提供的负荷数据进行聚类,并以Davies-Bouldin聚类有效性指标评估效果。结果表明该方法具有较好的划分能力,可以提高负荷曲线聚类的准确性。

甜仁杏优株果实性状调查与分析

对38个甜仁杏优株的单果重、单核重、出核率、单仁重、核壳厚度、出仁率、核长度、核宽度、核形指数、单株果核产量10个性状进行聚类分析、相关性分析和主成分分析。结果表明:当欧氏距离为8时,38个甜仁杏优株分为5类,其中第Ⅰ类为大果类型,但出核率和核形指数的平均值最低;第Ⅱ类单果重、单核重的平均值较大,但出核率和出仁率平均值较小,单株果核产量平均值较低;第Ⅲ类各项指标平均值较为居中,但出仁率平均值最低;第Ⅳ类各项指标平均值较为居中,且出仁率相对较高;第Ⅴ类出核率、出仁率、核形指数均最大,但其他各项指标均最小。单核重、单仁重均与核壳厚度、核长度、核宽度呈极显著正相关;出核率与单果重呈极显著负相关;出仁率与核壳厚度呈极显著负相关,与单核重呈显著负相关;单株果核产量与核长度、核宽度呈极显著正相关。主成分分析中,前4个主成分的累计方差贡献率达到83.949%,在得分排名前10名的优株中,果仁味道同时被评为优级的分别为13-32(一)、09-10(四)、16-46(一)、10-46(一)、10-03(二)、01-04(一),综合出仁率和果肉厚度考虑13-32(一)、09-10(四)、16-46(一)、10-46(一)将作为新品种选育的重点备选优株,进入新品种选育的下一个环节。