Nonnegative matrix factorization (NMF) is a method to get parts-based features of information and form the typical profiles. But the basis vectors NMF gets are not orthogonal so that parts-based features of informatio...Nonnegative matrix factorization (NMF) is a method to get parts-based features of information and form the typical profiles. But the basis vectors NMF gets are not orthogonal so that parts-based features of information are usually redundancy. In this paper, we propose two different approaches based on localized non-negative matrix factorization (LNMF) to obtain the typical user session profiles and typical semantic profiles of junk mails. The LNMF get basis vectors as orthogonal as possible so that it can get accurate profiles. The experiments show that the approach based on LNMF can obtain better profiles than the approach based on NMF. Key words localized non-negative matrix factorization - profile - log mining - mail filtering CLC number TP 391 Foundation item: Supported by the National Natural Science Foundation of China (60373066, 60303024), National Grand Fundamental Research 973 Program of China (2002CB312000), National Research Foundation for the Doctoral Program of Higher Education of China (20020286004).Biography: Jiang Ji-xiang (1980-), male, Master candidate, research direction: data mining, knowledge representation on the Web.展开更多
In inductively coupled plasma mass spectrometry (ICP-MS) analysis, only a few options are available to deal with non-spectroscopic interferences. Considering that diluting the sample is impractical for traces analysis...In inductively coupled plasma mass spectrometry (ICP-MS) analysis, only a few options are available to deal with non-spectroscopic interferences. Considering that diluting the sample is impractical for traces analysis, other alternatives must be employed. Traditionally, the method of standard additions is used to correct the matrix effect but it is a time consuming method. Others methods involves separation techniques. Another way to overcome matrix interferences is to understand the mechanism involved and adjust plasma viewing conditions to reduce or eliminate the effect. In this study, the effect of various concomitant elements in ICP-MS was assessed by measuring the distribution of selected singly charged analyte ions (Al, V, Cr, Mn, Ni, Co, Cu, Zn, As, In, Ba, La, Ce, Pb), doubly charged ions (La, Ce, Ba and Pb) and oxides ions (BaO) in the presence of concomitant elements spanning a mass range from 23 (Na) to 133 (Cs) u.m.a. and different ionization energies. Concomitant elements are alkali metals, alkaline earth metals and Si. Analyte ion suppression was observed while moving the ICP across and away from the sampling interface with or without a single concomitant element. Matrix effect measures were realised, firstly, to highlight the relation between the signal extinction of an analyte and the masse of the concomitant element, and secondly to highlight the relation between the removal of the analyte signal and the first ionization energy of the element of matrix. A dependence upon both the mass of the matrix element and the mass of the analyte was observed. The suppression seems increased with increasing matrix element mass and decreased with increasing analyte mass. The effect of the mass of the matrix element was the more significant of the two factors. If space-charge effects were found to be significant for matrix elements of much lower mass, it seems diffusion also played an active part for heavier matrix elements. Finally, some evidence was found for a shift in ion-atom equilibrium for dications and for energy demand regarding oxides.展开更多
The multimodal admittance method and its improvement are presented to deal with various aspects in underwater acoustics,mostly for the sound propagation in inhomogeneous waveguides with sound-speed profiles,arbitrary-...The multimodal admittance method and its improvement are presented to deal with various aspects in underwater acoustics,mostly for the sound propagation in inhomogeneous waveguides with sound-speed profiles,arbitrary-shaped liquid-like scatterers,and range-dependent environments.In all cases,the propagation problem governed by the Helmholtz equation is transformed into initial value problems of two coupled first-order evolution equations with respect to the modal components of field quantities(sound pressure and its derivative),by projecting the Helmholtz equation on a constructed orthogonal and complete local basis.The admittance matrix,which is the modal representation of Direchlet-to-Neumann operator,is introduced to compute the first-order evolution equations with no numerical instability caused by evanescent modes.The fourth-order Magnus scheme is used for the numerical integration of differential equations in the numerical implementation.The numerical experiments of sound field in underwater inhomogeneous waveguides generated by point sources are performed.Besides,the numerical results computed by simulation software COMSOL Multiphysics are given to validate the correction of the multimodal admittance method.It is shown that the multimodal admittance method is an efficient and stable numerical method to solve the wave propagation problem in inhomogeneous underwater waveguides with sound-speed profiles,liquid-like scatterers,and range-dependent environments.The extension of the method to more complicated waveguides such as horizontally stratified waveguides is available.展开更多
An efficient finite element model of involute helical gear is presented. A program based on compliance matrix method is developed for the calculation of instantaneous meshing stiffness, and the relationship between th...An efficient finite element model of involute helical gear is presented. A program based on compliance matrix method is developed for the calculation of instantaneous meshing stiffness, and the relationship between the rate of meshing stiffness and overlap ratio is given. On the basis of stiffness calculation, an optimization program for the optimal design of profile modification is developed according to the principle of internal point punishment function method.展开更多
A new electron density profile reconstruction procedure based on the PARK-matrix method has been firstly exploited for the multi-chord formic acid（HCOOH, λ=432.5 μm） laser interferometry system on the HL-2A tokama...A new electron density profile reconstruction procedure based on the PARK-matrix method has been firstly exploited for the multi-chord formic acid（HCOOH, λ=432.5 μm） laser interferometry system on the HL-2A tokamak. According to the geometric coordinates of the magnetic surfaces reconstructed by the CF（current fitting） code and the assumption that the electron density between two adjacent magnetic surfaces is a constant, the local electron density is calculated layer by layer, and the electron density profile n_e（Z） can be determined, as well as the density profile n_e（R）. The simulation result indicates that the error of the PARK-matrix method is acceptable for the four-chord HCOOH laser interferometer. In the applications, it shows that the reconstructed electron density profile agrees well with the microwave reflectometry measurement,and the sawtooth reversion radius is consistent with that deduced from the soft X-ray signals.Meanwhile, the electron density profiles with electron cyclotron resonance heating（ECRH） and supersonic molecular beam injection（SMBI） are also reconstructed and analyzed.展开更多
文摘Nonnegative matrix factorization (NMF) is a method to get parts-based features of information and form the typical profiles. But the basis vectors NMF gets are not orthogonal so that parts-based features of information are usually redundancy. In this paper, we propose two different approaches based on localized non-negative matrix factorization (LNMF) to obtain the typical user session profiles and typical semantic profiles of junk mails. The LNMF get basis vectors as orthogonal as possible so that it can get accurate profiles. The experiments show that the approach based on LNMF can obtain better profiles than the approach based on NMF. Key words localized non-negative matrix factorization - profile - log mining - mail filtering CLC number TP 391 Foundation item: Supported by the National Natural Science Foundation of China (60373066, 60303024), National Grand Fundamental Research 973 Program of China (2002CB312000), National Research Foundation for the Doctoral Program of Higher Education of China (20020286004).Biography: Jiang Ji-xiang (1980-), male, Master candidate, research direction: data mining, knowledge representation on the Web.
文摘In inductively coupled plasma mass spectrometry (ICP-MS) analysis, only a few options are available to deal with non-spectroscopic interferences. Considering that diluting the sample is impractical for traces analysis, other alternatives must be employed. Traditionally, the method of standard additions is used to correct the matrix effect but it is a time consuming method. Others methods involves separation techniques. Another way to overcome matrix interferences is to understand the mechanism involved and adjust plasma viewing conditions to reduce or eliminate the effect. In this study, the effect of various concomitant elements in ICP-MS was assessed by measuring the distribution of selected singly charged analyte ions (Al, V, Cr, Mn, Ni, Co, Cu, Zn, As, In, Ba, La, Ce, Pb), doubly charged ions (La, Ce, Ba and Pb) and oxides ions (BaO) in the presence of concomitant elements spanning a mass range from 23 (Na) to 133 (Cs) u.m.a. and different ionization energies. Concomitant elements are alkali metals, alkaline earth metals and Si. Analyte ion suppression was observed while moving the ICP across and away from the sampling interface with or without a single concomitant element. Matrix effect measures were realised, firstly, to highlight the relation between the signal extinction of an analyte and the masse of the concomitant element, and secondly to highlight the relation between the removal of the analyte signal and the first ionization energy of the element of matrix. A dependence upon both the mass of the matrix element and the mass of the analyte was observed. The suppression seems increased with increasing matrix element mass and decreased with increasing analyte mass. The effect of the mass of the matrix element was the more significant of the two factors. If space-charge effects were found to be significant for matrix elements of much lower mass, it seems diffusion also played an active part for heavier matrix elements. Finally, some evidence was found for a shift in ion-atom equilibrium for dications and for energy demand regarding oxides.
文摘The multimodal admittance method and its improvement are presented to deal with various aspects in underwater acoustics,mostly for the sound propagation in inhomogeneous waveguides with sound-speed profiles,arbitrary-shaped liquid-like scatterers,and range-dependent environments.In all cases,the propagation problem governed by the Helmholtz equation is transformed into initial value problems of two coupled first-order evolution equations with respect to the modal components of field quantities(sound pressure and its derivative),by projecting the Helmholtz equation on a constructed orthogonal and complete local basis.The admittance matrix,which is the modal representation of Direchlet-to-Neumann operator,is introduced to compute the first-order evolution equations with no numerical instability caused by evanescent modes.The fourth-order Magnus scheme is used for the numerical integration of differential equations in the numerical implementation.The numerical experiments of sound field in underwater inhomogeneous waveguides generated by point sources are performed.Besides,the numerical results computed by simulation software COMSOL Multiphysics are given to validate the correction of the multimodal admittance method.It is shown that the multimodal admittance method is an efficient and stable numerical method to solve the wave propagation problem in inhomogeneous underwater waveguides with sound-speed profiles,liquid-like scatterers,and range-dependent environments.The extension of the method to more complicated waveguides such as horizontally stratified waveguides is available.
文摘An efficient finite element model of involute helical gear is presented. A program based on compliance matrix method is developed for the calculation of instantaneous meshing stiffness, and the relationship between the rate of meshing stiffness and overlap ratio is given. On the basis of stiffness calculation, an optimization program for the optimal design of profile modification is developed according to the principle of internal point punishment function method.
基金supported by the National Magnetic Confinement Fusion Science Program of China(No.2014GB109001)National Natural Science Foundation of China(Nos.11505053 and 11275059)
文摘A new electron density profile reconstruction procedure based on the PARK-matrix method has been firstly exploited for the multi-chord formic acid（HCOOH, λ=432.5 μm） laser interferometry system on the HL-2A tokamak. According to the geometric coordinates of the magnetic surfaces reconstructed by the CF（current fitting） code and the assumption that the electron density between two adjacent magnetic surfaces is a constant, the local electron density is calculated layer by layer, and the electron density profile n_e（Z） can be determined, as well as the density profile n_e（R）. The simulation result indicates that the error of the PARK-matrix method is acceptable for the four-chord HCOOH laser interferometer. In the applications, it shows that the reconstructed electron density profile agrees well with the microwave reflectometry measurement,and the sawtooth reversion radius is consistent with that deduced from the soft X-ray signals.Meanwhile, the electron density profiles with electron cyclotron resonance heating（ECRH） and supersonic molecular beam injection（SMBI） are also reconstructed and analyzed.
