Feature selection by interactive clustering

The results of experiments with a novel criterion for absolute non-parametric feature selection are reported. The basic idea of the new technique involves the use of computer graphics and the human pattern recognition ability to interactively choose a number of features, this number not being necessarily determined in advance, from a larger set of measurements. The triangulation method, recently proposed in the cluster analysis literature for mapping points from l-space to 2-space, is used to yield a simple and efficient algorithm for feature selection by interactive clustering. It is shown that a subset of features can thus be chosen which allows a significant reduction in storage and time while still keeping the probability of error in classification within reasonable bounds.     

一种基于密度树的网格快速聚类算法的研究

聚类算法是数据挖掘领域中一个非常重要的研究方向。人们已经提出了许多适用于大规模的、高维的数据库的聚类算法。基于密度的聚类算法是其中一个比较典型的研究方向。该文以CABDET算法为基础，提出了一种基于密度树的网格快速聚类算法，该算法将网格的原理运用到基于密度树的聚类算法中，有效地提高了聚类的效率，降低了I/O的开销。     

HPLC-ELSD analysis of 18 platycosides from balloon flower roots (Platycodi Radix) sourced from various regions in Korea and geographical clustering of the cultivation areas

An effective HPLC method to analyse platycosides from the balloon flower root was developed using ELSD. The optimum resolution of the platycosides was achieved on an ODS column with gradient elution of eluent A, 30 mM ammonium acetate buffer (pH 4.81): methanol: acetonitrile = 75:5:20 (v/v/v), and B, 69:5:26 (v/v/v). Amongst 18 platycosides, platycoside E showed the highest content, followed by polygalacin D₂ and 3″-O-acetylplatyconic acid A. The sum of these three compounds was recommended for quality control of balloon flower root for medicinal purposes. The samples could be clustered into groups based on platycoside content. Group I, characterised by a high concentration of platycosides, was located near the west coast of Korea, whereas group II, characterised by a low concentration of platycosides, was located inland or in mountainous area. The method could be used to control the quality of balloon flower root.     

Bayesian Nonparametric Joint Mixture Model for Clustering Spatially Correlated Time Series

Abstract

We develop a Bayesian nonparametric joint mixture model for clustering spatially correlated time series based on both spatial and temporal similarities. In the temporal perspective, the pattern of a time series is flexibly modeled as a mixture of Gaussian processes, with a Dirichlet process (DP) prior over mixture components. In the spatial perspective, the spatial location is incorporated as a feature for clustering, like a time series being incorporated as a feature. Namely, we model the spatial distribution of each cluster as a DP Gaussian mixture density. For the proposed model, the number of clusters does not need to be specified in advance, but rather is automatically determined during the clustering procedure. Moreover, the spatial distribution of each cluster can be flexibly modeled with multiple modes, without determining the number of modes or specifying spatial neighborhood structures in advance. Variational inference is employed for the efficient posterior computation of the proposed model. We validate the proposed model using simulated and real-data examples. Supplementary materials for the article are available online.     

Fertigung einer universellen Hüftprothesenpfanne mittels Hochdruckblechumformung

In recent years, the use of hip prostheses has become a routine procedure. The use of total hip replacement has evolved in recent years to a routine procedure. Despite this experience, it always comes back to complications. Especially the migration or loosening of the acetabular component because of the artificial load adaptive bone remodeling is still a current problem. This is due to the changing mechanical situation after the implantation of the prosthesis. Another problem is the high bone loss during implantation, which complicates a revision of the prosthesis. One solution is the use of patient‐specific prostheses. So far, the use of these prostheses is limited due to the time‐consuming and cost‐intensive production. The overall objective of the presented project in this publication is the development and establishment of an innovative concept for the production of patient‐specific hip prosthesis made of titanium plates by sheet metal forming. The idea of this concept involves the development of a two‐stage process. First of all a standardized hip prosthesis is made by high‐pressure sheet metal forming and then individually in the second step. This publication contains the derivation of the standard prosthesis. For this purpose, a design methodology was generated, based on the so‐called agglomerative clustering. In addition, presents the results of the numerical preliminary design of the first stage of the process and the tool developed concept in the course of this paper.     

一种基于语言概念空间聚类的信息检索方法

提出了一种以语言概念空间中的概念为聚类对象的信息检索方法以及适合于该方法的聚类算法。该聚类算法通过曲线拟合技术来实现文本的自动阈值确定和聚类划分,并最终通过聚类间的迭代和结果修正来完成整个聚类过程。概念的引入为解决词语的同义、多义问题提供了有力保障。实验表明,采用该方法的信息检索系统,与Jelinek-Mercer、k-means模型相比有较高的准确率和召回率,效果理想。     

Nonhypothesis Analysis of a Mutagenic Soybean (Glycine max [L.]) Population for Protein and Fatty‐Acid Composition

Soybean is a major source of oil for food, feed, and biofuel production. Mutagenesis is a tool for creating unique traits useful in breeding programs. The aim of this study is to use nonhypothesis statistical testing methods to make decisions about a mutagenic population. To this end, a total of 1037 mutation lines and 28 wild‐type lines were analyzed for fatty‐acid composition and protein content. Principal component analysis (PCA) was used to analyze the fatty acid profile, multivariate analysis of variance (MANOVA) to build a selection model for seed weight per plant and weight per 10 seeds, and clustering in conjunction with power analysis to determine the minimum number of individuals needed to create a MANOVA selection model for the oil to protein content. Five of the 35 possible entries were identified by PCA analysis for stearic acid and four of 16 possible entries for oleic acid. Interestingly, most of the selected mutants were validated genetically. In fact, selected mutants with high seed stearic acid or high seed oleic acid contents were verified to carry mutations on GmFAD2‐1A, GmFAD2‐1B, and GmSACPD‐C genes. This shows a promising method of identifying smaller portion of the population to screen for desired mutations.     

应用灰色聚类法进行钻头优选研究

根据新疆油田夏子街组砂砾岩、泥质粉砂岩的钻头钻速低、钻头选型难的情况,针对已使用的钻头进行统计归纳后,使用灰色聚类法对该地层的钻头进行钻头优选,主要应用是将夏子街组使用过的钻头作为聚类对象集,根据钻头进尺、机械钻速、纯钻时间、钻头成本等作为聚类标准集,求出其聚类系数,并通过聚类系数大小比较钻头优劣。在优选出最优钻头以后,再将该地层使用过的另外一只钻头以及与最优钻头同一型号的钻头等效果进行对比,证明其合理性。通过新疆油田夏子街组上资料表明,该方法优选出来的钻头能有效提高钻速。     

Numerical computation for clustering of carbon particles with various sizes under both natural and magnetizing convections

The behavior of carbon particles from 0.01 to diameters in natural convection flow field of air was numerically studied including the effect of the magnetic field and the Brownian motion. One thousand carbon particles were released randomly in a vertical cylindrical enclosure whose height is equal to its radius. The enclosure was heated from below and cooled from above to produce the bulk flow of the natural convection of air in the atmospheric condition. A coil with electric current was set coaxially at the enclosure bottom to produce a magnetic field. Sample computations are carried out for the enclosure of 3 cm high and maximum magnetic field is 2.8 T or less. The results show that the natural convection of air is enhanced by the magnetic field. With the increase in the magnetic strength, more particles go along the fluid streak lines. With the increase in the particle size, more particles get together at the vortex center and cluster.