一种基于非线性特征提取的数据降维算法

为了获得良好的图像分类效果,需要采集尽可能多的图像数据特征,进而使得图像原始特征空间的维数越来越高,造成维数灾难.特征提取是通过线性或者非线性映射,将高维特征空间映射到低维空间,从而降低数据维数.现有的特征提取算法往往忽视了数据本身特有的复杂结构以及非线性因素,造成映射方向的模糊以及分类精确度的缺失.充分考虑了图像数据本身的二维特性,通过改进的非线性特征提取方法、流形学习方法来提取图像特征.实验表明,该算法在不影响图像分类效果的前提下可以大大降低数据维数,减少计算复杂度.

集中式多雷达系统维数不匹配处理技术

在集中式多雷达跟踪系统中,经常存在雷达的观测信息维数不一致的情况。为了能正确对所有雷达观测数据进行融合,系统需要首先对维数不匹配的观测数据进行处理。目前,通常采用降维的方法处理数据维数不匹配问题。由于在降维处理过程中,需要丢失目标的高度信息,因此,系统会产生一定的动态误差。当目标的俯仰角度比较大时,系统的跟踪精度必然会下降。针对降维方法的缺点,本文采用升维的方法对数据维数不匹配进行处理,并对这两种方法在不同情况下的跟踪精度进行了仿真分析。

一种基于潜在语义线性嵌入的三维模型检索方法

在三维模型检索中,为了保持模型之间的形状相似性,并且考虑保持模型之间的语义关系不变的情况下实现特征的维数缩减,提出一个基于潜在语义分析的非线性降维方法-潜在语义线性嵌入LSLE.该方法通过构建的三维模型语义空间确定数据库中三维模型的语义近邻,并通过改进的局部线性嵌入方法实现数据维数的缩减,对Princeton Shape Benchmark的实验表明,该方法在三维模型的特征降维和检索中都取得了较好的效果.

基于粗集和支持向量机的雷达信号识别法

结合粗集的属性约简和支持向量机的分类机理,提出了一种融合算法。应用粗集理论的属性约简过程作为数据的预处理,把冗余的属性和冲突的对象从决策表中删去,却不损失任何有效信息;然后利用支持向量机进行分类识别。这样可以大大降低数据维数,减小支持向量机分类过程中的复杂度,在不同程度上避免了训练模型的过拟合现象,且分类性能更加优越。最后通过对雷达信号识别的仿真实例说明了上述方法的有效性。

基于蚁群优化的特征基因选择算法

传统的特征选择算法需要依靠大量的数据进行有监督训练,具有高维和小样本的属性,进而造成数据冗余,导致维数灾难。针对这一情况,提出一种基于蚁群优化的特征基因选择算法。该算法对传统的蚁群算法模型进行优化,改进蚁群算法的参数选择方法,利用特征对不同数据集的敏感度,寻找最优基因,滤除无关基因,将特征选择过程转化为蚁群寻找最优路径的过程。实验表明,该算法可以有效地对特征进行优化选择,在降低数据维数的同时,提高分类的准确性和时效性。

Experimental and Kinetic Study of Selective Catalytic Reduction of NO with NH_3 over CuO/Al_2O_3/Cordierite Catalyst

The CuO/γ-Al2O3/cordierite catalyst, after being sulfated by sulfur dioxide (SO2) at 673 K, exhibits high activities for selective catalytic reduction (SCR) of nitrogen oxide (NO) with ammonia (NH3) at 573-723 K. The intrinsic kinetics of SCR of NO with NH3 over CuO/γ-Al2O3/cordierite catalyst has been measured in a fixed-bed reactor in the absence of internal and external diffusions. The experimental results show that the reaction rate can be quantified by a first-order expression with activation energy Eá of 94.01 kJ·mol-1 and the corresponding p re-exponential factor A′ of 3.39×108 cm3·g-1·s-1 when NH3 is excessive. However, when NH3 is not enough, an E ley-Rideal kinetic model based on experimental data is derived with Ea of 105.79 kJ·mol-1, the corresponding A of 2 .94×109 cm3·g-1·s-1, heat of adsorption-Hads of 87.90 kJ·mol-1 and the corresponding Aads of 9.24 cm3·mol-1. The intrinsic kinetic model obtained was incorporated in a 3D mathematical model of monolithic reactor, and the agreement of the prediction with experimental data indicates that the present kinetic model is adequate for the reac-tor design and engineering scale-up.

Transect-based Three-Dimensional Road Modeling and Visualization

On the basis of the study on road elements and th e existing three-dimensional road data models, this paper puts forward the transe ct-based road modeling, resolves the modeling of the simplest road with two roa d transects, and fulfils the visualization of the road. OpenGL is applied as the tool of visualization. The map texture technology, LOD algorithm and dynamic mu lti-differentiating texture technology are of benefit to the construction of 3D road GIS, and improve the speed of roaming and effective visualization. This pa per also disusses the combination of 2D digitall road map and 3D road scene, and the dynamic response between them. Finally, on the basis of the research on the transect-based road model, this paper develops the three-dimensional road geo graphic information system called virtual road ( VRoad ), which not only suppli es road designers with a set of tool which can turn the designed 2D road data in to 3D road and the high road assistant function area in computer, but also suppl ies the road management with a set of tool which can realize the road real time and interactive roaming, high-efficiency management.

Similarity measure design for high dimensional data

Information analysis of high dimensional data was carried out through similarity measure application. High dimensional data were considered as the a typical structure. Additionally, overlapped and non-overlapped data were introduced, and similarity measure analysis was also illustrated and compared with conventional similarity measure. As a result, overlapped data comparison was possible to present similarity with conventional similarity measure. Non-overlapped data similarity analysis provided the clue to solve the similarity of high dimensional data. Considering high dimensional data analysis was designed with consideration of neighborhoods information. Conservative and strict solutions were proposed. Proposed similarity measure was applied to express financial fraud among multi dimensional datasets. In illustrative example, financial fraud similarity with respect to age, gender, qualification and job was presented. And with the proposed similarity measure, high dimensional personal data were calculated to evaluate how similar to the financial fraud. Calculation results show that the actual fraud has rather high similarity measure compared to the average, from minimal 0.0609 to maximal 0.1667.

Linear manifold clustering for high dimensional data based on line manifold searching and fusing

High dimensional data clustering,with the inherent sparsity of data and the existence of noise,is a serious challenge for clustering algorithms.A new linear manifold clustering method was proposed to address this problem.The basic idea was to search the line manifold clusters hidden in datasets,and then fuse some of the line manifold clusters to construct higher dimensional manifold clusters.The orthogonal distance and the tangent distance were considered together as the linear manifold distance metrics. Spatial neighbor information was fully utilized to construct the original line manifold and optimize line manifolds during the line manifold cluster searching procedure.The results obtained from experiments over real and synthetic data sets demonstrate the superiority of the proposed method over some competing clustering methods in terms of accuracy and computation time.The proposed method is able to obtain high clustering accuracy for various data sets with different sizes,manifold dimensions and noise ratios,which confirms the anti-noise capability and high clustering accuracy of the proposed method for high dimensional data.

Three-dimensional conjugate gradient inversion of magnetotelluric full information data

Based on the analysis of impedance tensor data, tipper data, and the conjugate gradient algorithm, we develop a three-dimensional （3D） conjugate gradient algorithm for inverting magnetotelluric full information data determined from five electric and magnetic field components and discuss the method to use the full information data for quantitative interpretation of 3D inversion results. Results from the 3D inversion of synthetic data indicate that the results from inverting full information data which combine the impedance tensor and tipper data are better than results from inverting only the impedance tensor data （or tipper data） in improving resolution and reliability. The synthetic examples also demonstrate the validity and stability of this 3D inversion algorithm.

Differentially Private Multidimensional Data Publication

Multidimensional data provides enormous opportunities in a variety of applications. Recent research has indicated the failure of existing sanitization techniques （e.g., k-anonymity） to provide rigorous privacy guarantees. Privacy- preserving multidimensional data publishing currently lacks a solid theoretical foundation. It is urgent to develop new techniques with provable privacy guarantees, e-Differential privacy is the only method that can provide such guarantees. In this paper, we propose a multidimensional data publishing scheme that ensures c-differential privacy while providing accurate results for query processing. The proposed solution applies nonstandard wavelet transforms on the raw multidimensional data and adds noise to guarantee c-differential privacy. Then, the scheme processes arbitrarily queries directly in the noisy wavelet- coefficient synopses of relational tables and expands the noisy wavelet coefficients back into noisy relational tuples until the end result of the query. Moreover, experimental results demonstrate the high accuracy and effectiveness of our approach.

Research on three-dimension ocean observation data integration and service technology

Currently,ocean data portals are being developed around the world based on Geographic Information Systems(GIS) as a source of ocean data and information.However,given the relatively high temporal frequency and the intrinsic spatial nature of ocean data and information,no current GIS software is adequate to deal effectively and efficiently with spatiotemporal data.Furthermore,while existing ocean data portals are generally designed to meet the basic needs of a broad range of users,they are sometimes very complicated for general audiences,especially for those without training in GIS.In this paper,a new technical architecture for an ocean data integration and service system is put forward that consists of four layers:the operation layer,the extract,transform,and load(ETL) layer,the data warehouse layer,and the presentation layer.The integration technology based on the XML,ontology,and spatiotemporal data organization scheme for the data warehouse layer is then discussed.In addition,the ocean observing data service technology realized in the presentation layer is also discussed in detail,including the development of the web portal and ocean data sharing platform.The application on the Taiwan Strait shows that the technology studied in this paper can facilitate sharing,access,and use of ocean observation data.The paper is based on an ongoing research project for the development of an ocean observing information system for the Taiwan Strait that will facilitate the prevention of ocean disasters.

Junk band recovery for hyperspectral image based on curvelet transform

Under consideration that the profiles of bands at close wavelengths are quite similar and the curvelets are good at capturing profiles, a junk band recovery algorithm for hyperspectral data based on curvelet transform is proposed. Both the noisy bands and the noise-free bands are transformed via curvelet band by band. The high frequency coefficients in junk bands are replaced with linear interpolation of the high frequency coefficients in noise-flee bands, and the low frequency coefficients remain the same to keep the main spectral characteristics from being distorted. Jutak bands then are recovered after the inverse curvelet transform. The performance of this method is tested on the hyperspectral data cube obtained by airborne visible/infrared imaging spectrometer （AVIRIS）. The experimental results show that the proposed method is superior to the traditional denoising method BayesShrink and the art-of-state Curvelet Shrinkage in both roots of mean square error （RMSE） and peak-signal-to-noise ratio （PSNR） of recovered bands.

Application of principal component analysis to evaluation of black soil degradation in Jilin

Principal Component Analysis(PCA) can simplify the structure of database by replacing multi-dimensional parameters with relatively less comprehensive variables in order to ensure the minimum lost in initial data.In this paper,eighteen black soil samples from different sites were tested and thirteen distinctive indexes were chosen to evaluate the degeneration of black soil.By using principal component analysis,variables of thirteen dimensions can be diminished to six unrelated principal indexes.Analysis shows that the soluble salt content,Fulvic acids(FA) and aggregation degree have a high weighing coefficient,indicating these three indexes are the major parts for the evaluation of black soil degradation.It also provides a new path to the degenerated black soil treatment in Northeast China.

The development of real time data driving multi-axis linkage and synergic movement control system of 3D variable cross-section roll forming machine

The three dimensional variable cross-section roll forming is a kind of new metal forming technol- ogy which combines large forming force, multi-axis linkage movement and space synergic movement, and the sequential synergic movement of the ganged roller group is used to complete the metal sheet forming according to the shape of the complicated and variable forming part data. The control system should meet the demands of quick response to the test requirements of the product part. A new kind of real time data driving multi-axis linkage and synergic movement control strategy of 3D roll forming is put forward in the paper. In the new control strategy, the forming data are automatically generated according to the shape of the parts, and the multi-axis linkage movement together with cooperative motion among the six stands of the 3D roll forming machine is driven by the real-time information, and the control nodes are also driven by the forming data. The new control strategy is applied to a 48 axis 3D roll forming machine developed by our research center, and the control servo period is less than 10ms. A forming experiment of variable cross section part is carried out, and the forming preci- sion is better than ＋ 0.5mm by the control strategy. The result of the experiment proves that the control strategy has significant potentiality for the development of 3D roll forming production line with large scale, multi-axis ganged and svner^ic movement

Research on Multidimensional Data Model in Data Warehouse

Data structure and semantics of the traditional data model cannot effectively represent the data warehouse, it is difficult to effectively support online analytical processing （referred to as OLAP）. This paper is propose a new multidimensional data model based on the partial ordering and mapping. The data model can fully express the complex data structure and semantics of data warehouse, and provide an OLAP operation as the core of the operation of algebra, support structure in levels of complex aggregation operation sequence, which can effectively support the application of OLAE The data model supports the concept of aggregation function constraint, and provides constraint mechanism of the hierarchy aggregation function.

The Proposal of Data Warehouse Validation

The analysis of relevant standards and guidelines proved the lack of information on actions and activities concerning data warehouse testing. The absence of the complex data warehouse testing methodology seems to be crucial particularly in the phase of the data warehouse implementation. The aim of this article is to suggest basic data warehouse testing activities as a final part of data warehouse testing methodology. The testing activities that must be implemented in the process of the data warehouse testing can be split into four logical units regarding the multidimensional database testing, data pump testing, metadata and OLAP （Online Analytical Processing） testing. Between main testing activities can be included： revision of the multidimensional database scheme, optimizing of fact tables number, problem of data explosion, testing for correctness of aggregation and summation of data etc.

GPU Accelerated Marine Data Visualization Method

The study of marine data visualization is of great value. Marine data, due to its large scale, random variation and multiresolution in nature, are hard to be visualized and analyzed. Nowadays, constructing an ocean model and visualizing model results have become some of the most important research topics of ‘Digital Ocean'. In this paper, a spherical ray casting method is developed to improve the traditional ray-casting algorithm and to make efficient use of GPUs. Aiming at the ocean current data, a 3D view-dependent line integral convolution method is used, in which the spatial frequency is adapted according to the distance from a camera. The study is based on a 3D virtual reality and visualization engine, namely the VV-Ocean. Some interactive operations are also provided to highlight the interesting structures and the characteristics of volumetric data. Finally, the marine data gathered in the East China Sea are displayed and analyzed. The results show that the method meets the requirements of real-time and interactive rendering.

Discriminant embedding by sparse representation and nonparametric discriminant analysis for face recognition

A novel supervised dimensionality reduction algorithm, named discriminant embedding by sparse representation and nonparametric discriminant analysis(DESN), was proposed for face recognition. Within the framework of DESN, the sparse local scatter and multi-class nonparametric between-class scatter were exploited for within-class compactness and between-class separability description, respectively. These descriptions, inspired by sparse representation theory and nonparametric technique, are more discriminative in dealing with complex-distributed data. Furthermore, DESN seeks for the optimal projection matrix by simultaneously maximizing the nonparametric between-class scatter and minimizing the sparse local scatter. The use of Fisher discriminant analysis further boosts the discriminating power of DESN. The proposed DESN was applied to data visualization and face recognition tasks, and was tested extensively on the Wine, ORL, Yale and Extended Yale B databases. Experimental results show that DESN is helpful to visualize the structure of high-dimensional data sets, and the average face recognition rate of DESN is about 9.4%, higher than that of other algorithms.