The support vector machine （SVM） is a novel machine learning method, which has the ability to approximate nonlinear functions with arbitrary accuracy. Setting parameters well is very crucial for SVM learning results...The support vector machine （SVM） is a novel machine learning method, which has the ability to approximate nonlinear functions with arbitrary accuracy. Setting parameters well is very crucial for SVM learning results and generalization ability, and now there is no systematic, general method for parameter selection. In this article, the SVM parameter selection for function approximation is regarded as a compound optimization problem and a mutative scale chaos optimization algorithm is employed to search for optimal paraxneter values. The chaos optimization algorithm is an effective way for global optimal and the mutative scale chaos algorithm could improve the search efficiency and accuracy. Several simulation examples show the sensitivity of the SVM parameters and demonstrate the superiority of this proposed method for nonlinear function approximation.展开更多
In this paper, the exact Bayesian limits, taking conjugate and noninformative prior distribution, and the exact fiducial limits for the mean of the lognormal distribution are presented. They can be found iteratively b...In this paper, the exact Bayesian limits, taking conjugate and noninformative prior distribution, and the exact fiducial limits for the mean of the lognormal distribution are presented. They can be found iteratively by one-dimension integral on a finite interval. The new algorithm is very convenient and with high accuracy. It can meet the practical engineering need excellently. However, the primitive algorithm is rather cumbersome. By the way, the very close approximate limits with a simple algorithm are derived. They can be applied immediately to engineering. Otherwise, they can also be used as a search interval to find the root of equation for the exact limits.展开更多
In this paper, a theoretical model of multi-level, non-spherical scatterers is developed for fully polarimetric scattering from tree canopy in SAR imaging at C band. The amplitude functions of non-spherical particles ...In this paper, a theoretical model of multi-level, non-spherical scatterers is developed for fully polarimetric scattering from tree canopy in SAR imaging at C band. The amplitude functions of non-spherical particles with randomly spatial orientation are derived by the generalized Rayleigh-Gans (GRG) approximation. The non-diagonal extinction matrix and the Mueller matrix solution are constructed. Numerical solutions of polarimetric scattering of four Stokes parameters from random, non-spherical scatterers are obtained. To physically identify polarimetric scattering of the Mueller matrix solution, the coherency matrix and its eigen-analysis are discussed. Functional dependence of the coherency matrix and entropy upon various parameters are obtained. As an application, the analysis of AirSAR images at P, L, C bands is discussed.展开更多
Wavelet has been used as a powerful tool in the signal processing and function approximation recently. This paper presents the application of wavelets for solving two key problems in 3-D audio simulation. First, we em...Wavelet has been used as a powerful tool in the signal processing and function approximation recently. This paper presents the application of wavelets for solving two key problems in 3-D audio simulation. First, we employ discrete wavelet transform (DWT) combined with vector quantization (VQ) to compress audio data in order to reduce tremendous redundant data storage and transmission times. Secondly, we use wavelets as the activation functions in neural networks called feed-forward wavelet networks to approach auditory localization information cues (head-related transfer functions (HRTFs) are used here). The experimental results demonstrate that the application of wavelets is more efficient and useful in 3-D audio simulation.展开更多
Local inhomogeneity in totally asymmetric simple exclusion processes （TASEPs） with different hopping rates was studied. Many biological and chemical phenomena can be described by these non-equilibrium processes. A s...Local inhomogeneity in totally asymmetric simple exclusion processes （TASEPs） with different hopping rates was studied. Many biological and chemical phenomena can be described by these non-equilibrium processes. A simple approximate theory and extensive Monte Carlo computer simulations were used to calculate the steady-state phase diagrams and bulk densities. It is found that the phase diagram for local inhomogeneity in TASEP with different hopping rates p is qualitatively similar to homogeneous models. Interestingly, there is a saturation point pair （a＊, fl＊） for the system, which is decided by parameters p and q. There are three stationary phases in the system, when parameter p is fixed （i.e., p=0.8）, with the increase of the parameter q, the region of LD/LD and HD/HD phase increases and the HD/LD is the only phase which the region shrinks. The analytical results are in good agreement with simulations.展开更多
基金the National Nature Science Foundation of China (60775047, 60402024)
文摘The support vector machine （SVM） is a novel machine learning method, which has the ability to approximate nonlinear functions with arbitrary accuracy. Setting parameters well is very crucial for SVM learning results and generalization ability, and now there is no systematic, general method for parameter selection. In this article, the SVM parameter selection for function approximation is regarded as a compound optimization problem and a mutative scale chaos optimization algorithm is employed to search for optimal paraxneter values. The chaos optimization algorithm is an effective way for global optimal and the mutative scale chaos algorithm could improve the search efficiency and accuracy. Several simulation examples show the sensitivity of the SVM parameters and demonstrate the superiority of this proposed method for nonlinear function approximation.
文摘In this paper, the exact Bayesian limits, taking conjugate and noninformative prior distribution, and the exact fiducial limits for the mean of the lognormal distribution are presented. They can be found iteratively by one-dimension integral on a finite interval. The new algorithm is very convenient and with high accuracy. It can meet the practical engineering need excellently. However, the primitive algorithm is rather cumbersome. By the way, the very close approximate limits with a simple algorithm are derived. They can be applied immediately to engineering. Otherwise, they can also be used as a search interval to find the root of equation for the exact limits.
文摘In this paper, a theoretical model of multi-level, non-spherical scatterers is developed for fully polarimetric scattering from tree canopy in SAR imaging at C band. The amplitude functions of non-spherical particles with randomly spatial orientation are derived by the generalized Rayleigh-Gans (GRG) approximation. The non-diagonal extinction matrix and the Mueller matrix solution are constructed. Numerical solutions of polarimetric scattering of four Stokes parameters from random, non-spherical scatterers are obtained. To physically identify polarimetric scattering of the Mueller matrix solution, the coherency matrix and its eigen-analysis are discussed. Functional dependence of the coherency matrix and entropy upon various parameters are obtained. As an application, the analysis of AirSAR images at P, L, C bands is discussed.
文摘Wavelet has been used as a powerful tool in the signal processing and function approximation recently. This paper presents the application of wavelets for solving two key problems in 3-D audio simulation. First, we employ discrete wavelet transform (DWT) combined with vector quantization (VQ) to compress audio data in order to reduce tremendous redundant data storage and transmission times. Secondly, we use wavelets as the activation functions in neural networks called feed-forward wavelet networks to approach auditory localization information cues (head-related transfer functions (HRTFs) are used here). The experimental results demonstrate that the application of wavelets is more efficient and useful in 3-D audio simulation.
基金Project(2011FZ050) supported by Applied Basic Research Program of Yunnan Provincial Science and Technology Department,ChinaProject(2011J084) supported by Master Program of Yunnan Province Education Department,China
文摘Local inhomogeneity in totally asymmetric simple exclusion processes （TASEPs） with different hopping rates was studied. Many biological and chemical phenomena can be described by these non-equilibrium processes. A simple approximate theory and extensive Monte Carlo computer simulations were used to calculate the steady-state phase diagrams and bulk densities. It is found that the phase diagram for local inhomogeneity in TASEP with different hopping rates p is qualitatively similar to homogeneous models. Interestingly, there is a saturation point pair （a＊, fl＊） for the system, which is decided by parameters p and q. There are three stationary phases in the system, when parameter p is fixed （i.e., p=0.8）, with the increase of the parameter q, the region of LD/LD and HD/HD phase increases and the HD/LD is the only phase which the region shrinks. The analytical results are in good agreement with simulations.
