新的文本分类特征选择方法研究

特征降维是文本分类过程中的一个重要环节。在现有特征选择方法的基础上,综合考虑特征词在正类和负类中的分布性质,综合四种衡量特征类别区分能力的指标,提出了一个新的特征选择方法,即综合比率(CR)方法。实验采用K-最近邻分类算法(KNN)来考查CR方法的有效性,实验结果表明该方法能够取得比现有特征选择方法更优的降维效果。     

基于遗传算法的C-Bézier曲线降阶

针对C-Bézier曲线的近似降阶问题,基于遗传算法,给出了一种用n次C-Bézier曲线最小平方逼近n+1次C-Bézier曲线的方法。该方法从最优化思想出发,把C-Bézier曲线的降阶问题转化为求解函数的优化问题,通过选择适应值函数,利用简单的循环执行复制、交叉、变异、选择求出该优化问题的最优值,从而实现了C-Bézier曲线在端点无约束和端点G0约束条件下的近似降阶逼近。实例结果表明,所提方法不仅可以获得较好的降阶效果,而且易于实现、精度高、误差计算简单,可以广泛地应用于计算机辅助设计中对曲线的近似降阶。     

涞源岩溶地下水系统泉水量变化特征分析

涞源泉泉域是一个全封闭式泉排型岩溶水系统,泉水利用对当地工农业生产发挥了重要作用.通过对涞源岩溶地下水系统特征分析,利用地表水系列资料,采用基流分割法、相关系数法等,计算出泉水量系列资料.由降水补给地下水,再以泉水的形式排泄,在此过程中使水量在时间上滞后,而且水量趋于平稳,泉水量年极值比为7.33,变差系数为0.34,年内分配变化幅度平缓.由降水—入渗—地下径流—泉水排出的过程,水资源经过地下岩溶调节作用,有利于水资源的开发利用,而且对水质也起到净化作用.研究和分析岩溶地下水径流量变化特征,为合理利用泉水资源提供科学依据.     

Identification and data-driven model reduction of state-space representations of lossless and dissipative systems from noise-free data

We illustrate procedures to identify a state-space representation of a lossless or dissipative system from a given noise-free trajectory; important special cases are passive systems and bounded-real systems. Computing a rank-revealing factorization of a Gramian-like matrix constructed from the data, a state sequence can be obtained; the state-space equations are then computed by solving a system of linear equations. This idea is also applied to perform model reduction by obtaining a balanced realization directly from data and truncating it to obtain a reduced-order model.     

Efficient parallel implementation of Ewald summation in molecular dynamics simulations on multi-core platforms

We present a multi-step computation method to implement the Ewald summation for long-range electrostatic interactions in molecular dynamic simulations on a multi-core machine. Our methodology is based on the OpenMP programming model. It partitions computations of real-space summation among threads so that the global force of a single particle cannot be modified by more than one thread simultaneously. It requires neither a private copy of the force array for each thread nor an inspector at runtime. Compared with some other methods that can parallelise reduction operations on a force-array, our method achieves relatively higher speedups and lower L2 cache miss and bus utilisation ratios.     

A time-efficient pattern reduction algorithm for k-means clustering

This paper presents an efficient algorithm, called pattern reduction (PR), for reducing the computation time of k-means and k-means-based clustering algorithms. The proposed algorithm works by compressing and removing at each iteration patterns that are unlikely to change their membership thereafter. Not only is the proposed algorithm simple and easy to implement, but it can also be applied to many other iterative clustering algorithms such as kernel-based and population-based clustering algorithms. Our experiments—from 2 to 1000 dimensions and 150 to 10,000,000 patterns—indicate that with a small loss of quality, the proposed algorithm can significantly reduce the computation time of all state-of-the-art clustering algorithms evaluated in this paper, especially for large and high-dimensional data sets.     

Necessary optimality conditions for fuzzy variational problems

The main aims of this study are to derive the fuzzy Euler-Lagrange conditions for both fuzzy unconstrained and constrained variational problems based on the concepts of differentiability and integrability of a fuzzy mapping that may be parameterized by the left and right-hand functions of its α-level sets.     

Supervised principal component analysis: Visualization, classification and regression on subspaces and submanifolds

We propose “supervised principal component analysis (supervised PCA)”, a generalization of PCA that is uniquely effective for regression and classification problems with high-dimensional input data. It works by estimating a sequence of principal components that have maximal dependence on the response variable. The proposed supervised PCA is solvable in closed-form, and has a dual formulation that significantly reduces the computational complexity of problems in which the number of predictors greatly exceeds the number of observations (such as DNA microarray experiments). Furthermore, we show how the algorithm can be kernelized, which makes it applicable to non-linear dimensionality reduction tasks. Experimental results on various visualization, classification and regression problems show significant improvement over other supervised approaches both in accuracy and computational efficiency.     

A survey of multilinear subspace learning for tensor data

Increasingly large amount of multidimensional data are being generated on a daily basis in many applications. This leads to a strong demand for learning algorithms to extract useful information from these massive data. This paper surveys the field of multilinear subspace learning (MSL) for dimensionality reduction of multidimensional data directly from their tensorial representations. It discusses the central issues of MSL, including establishing the foundations of the field via multilinear projections, formulating a unifying MSL framework for systematic treatment of the problem, examining the algorithmic aspects of typical MSL solutions, and categorizing both unsupervised and supervised MSL algorithms into taxonomies. Lastly, the paper summarizes a wide range of MSL applications and concludes with perspectives on future research directions.     

From classifiers to discriminators: A nearest neighbor rule induced discriminant analysis

The current discriminant analysis method design is generally independent of classifiers, thus the connection between discriminant analysis methods and classifiers is loose. This paper provides a way to design discriminant analysis methods that are bound with classifiers. We begin with a local mean based nearest neighbor (LM-NN) classifier and use its decision rule to supervise the design of a discriminator. Therefore, the derived discriminator, called local mean based nearest neighbor discriminant analysis (LM-NNDA), matches the LM-NN classifier optimally in theory. In contrast to that LM-NNDA is a NN classifier induced discriminant analysis method, we further show that the classical Fisher linear discriminant analysis (FLDA) is a minimum distance classifier (i.e. nearest Class-mean classifier) induced discriminant analysis method. The proposed LM-NNDA method is evaluated using the CENPARMI handwritten numeral database, the NUST603 handwritten Chinese character database, the ETH80 object category database and the FERET face image database. The experimental results demonstrate the performance advantage of LM-NNDA over other feature extraction methods with respect to the LM-NN (or NN) classifier.