基于概率加权平均的Mel子带特征重建算法

本文提出基于概率加权平均的Mel子带特征数据重建算法.该算法选择K个最优重建结果的概率加权平均作为被加性噪声掩蔽的语音特征分量的估计.实验结果表明,基于概率加权平均的语音特征数据重建算法降低了重建误差,减少了帧间突变现象,增强了Mel子带特征的帧间连续性,从而显著提高了语音识别系统对加性噪声的鲁棒性能.     

机载大斜视SAR的运动补偿方法

对机载大斜视的合成孔径雷达成像因速度变化引起的运动补偿方面的问题展开了研究 ,提出了大斜视角条件下运动补偿的方法 ,对其分别进行一次和二次相位补偿 ,并在进行方位压缩时考虑三次相位 ,其结果能较好地实现运动补偿。同时对补偿性能也作了分析 ,对基于运动参数估计的窄波束宽幅SAR成像算法进行了改进 ,增加一次和二次相位补偿的步骤 ,成为适用于机载大斜视SAR的成像算法     

A simple procedure for evaluating the heterogeneity parameter and the ratio of molecular sizes of solute and solvent by using HPLC data

Based on the relationship defining the dependence of the capacity ratio upon the composition of the mobile phase, a simple numerical procedure is proposed for evaluating the heterogeneity parameter and the ratio of molecular sizes of solute and solvent. This procedure is examined by using HPLC data for pesticides chromatographed in isooctane/dichloromethane eluent on Lichrosorb at 291 K.     

On defining OLAP formulations

^** Email: vutsinas{at}upatras.gr Recently there has been increasing interest in On Line AnalyticalProcessing (OLAP) to satisfy the organizational needs of high-levelinformation delivery and advanced data analysis. The actualapplication of OLAP tools involves the use of various functions,such as the common drilling down and slicing and dicing. Usuallyeach particular OLAP function is comprehensive and intuitive.However, sophisticated use of OLAP tools requires complicatedcombinations of different OLAP functions that are not straight-forwardfor end users or designers. In this paper we attempt to enumerateand formally define OLAP functions by defining a new OLAP modelthat provides a broader view of OLAP. We demonstrate the expressiveadequacy of the new OLAP model with application examples.     

Linear low-rank approximation and nonlinear dimensionality reduction

We present our recent work on both linear and nonlinear data reduction methods and algorithms: for the linear case we discuss results on structure analysis of SVD of column-partitioned matrices and sparse low-rank approximation; for the nonlinear case we investigate methods for nonlinear dimensionality reduction and manifold learning. The problems we address have attracted great deal of interest in data mining and machine learning.     

Semilinear parabolic equations involving measures and low regularity data

A detailed study of abstract semilinear evolution equations of the form is undertaken, where generates an analytic semigroup and is a Banach space valued measure depending on the solution. Then it is shown that the general theorems apply to a variety of semilinear parabolic boundary value problems involving measures in the interior and on the boundary of the domain. These results extend far beyond the known results in this field. A particularly new feature is the fact that the measures may depend nonlinearly and possibly nonlocally on the solution.

     

Fast topological analysis of 2D and 3D grids of data: application to the atoms in molecule (AIM) and the electron localization function (ELF)

The topological analysis of grids of data is used for determination of surfaces or volumes around maxima. The volumes are then related to chemical information such as atoms or bonds, and can be used for integration of local properties such as electronic population. The problem of global connectivity is reversed into the question of local connectivity yielding a linear scaling partition algorithm. Two packages are developed for a very fast analysis and partition of 2D or 3D grids of data, applications being made to C2H2, C2H4, C6H6, H2CO, and H2CS molecules using the Atoms in Molecule (AIM) or Electron Localization Function (ELF).     

Adaptive Correction of Errors in Residue Coupled MASH ΔΣ Modulators

MASH delta-sigma () modulators consist of a cascade of several lower order single-loop modulators. In an ideal cascade, the quantization error from all but the last stage are digitally canceled. The drawback with a cascaded design is the requirement of precise matching of contributions from different quantizers to cancel lower order quantization noise from intermediate delta-sigma stages. This paper presents a new, adaptive improvement to the residue coupled MASH delta sigma modulator. The adaptive corrections significantly reduce the sensitivity to analog imperfections. The result is a simple MASH delta-sigma modulator with high precision. Simulations of a 1-1 MASH circuit structure with errors and corrections are included to confirm the theory.     

Network Assisted Power Control for Wireless Data

The cellular telephone success story prompts the wireless communications community to turn its attention to other information services, many of them in the category of wireless data communications. One lesson of cellular telephone network operation is that effective power control is essential to promote system quality and efficiency. In recent we have applied microeconomic theories to power control taking into account notions of utility and pricing. Our earlier work has shown that this new approach to power control for wireless data performs better than traditional techniques applied for voice signals. However, the operating points of such a strategy result in an unfair equilibrium in that users operate with unequal signal-to-interference ratios. Further, the power control algorithms required to achieve such operating points are more complex than the simple signal-to-interference ratio balancing algorithms for voice. In this paper, we introduce a new concept, Network Assisted Power Control (NAPC) that maximizes utilities for users while maintaining equal signal-to-interference ratios for all users. The power control algorithm is easily implemented via signal-to-interference ratio balancing with the assistance of the network that broadcasts the common signal-to-interference ratio target.     

Buckling of a Timoshenko beam on elastic foundation with uncertain input data

A monotone dependence of the critical buckling load of a simplysupported Timoshenko–Mindlin beam both on the shear correctionfactor and on the stiffness of foundation is proved. Then theworst-scenario method is employed to find the ‘most dangerous’input data.