| 基于TSVD正则化方法的概率密度估计 |
| |
| 引用本文： | 吴笛,刘文.基于TSVD正则化方法的概率密度估计[J].武汉工学院学报,2012(1):60-63. |
| |
| 作者姓名： | 吴笛 刘文 |
| |
| 作者单位： | 武汉理工大学理学院,湖北武汉430070 |
| |
| 基金项目： | 中央高校基本科研业务费专项资金资助项目（2010-ZY-LX-022和2010-ZY-LX-027） |
| |
| 摘 要： | 从概率密度的定义出发,将概率密度估计转化成线性算子方程的求解,根据算子方程核矩阵奇异值的性质,构建了概率密度估计的TSVD正则化方法,并与线性Bregman迭代正则化方法进行了比较分析。从仿真结果来看,TSVD能更好地逼近真实函数,在不同噪声水平下表现出更强的鲁棒性。
|
| 关 键 词： | 概率密度估计 截断奇异值分解 Bregman迭代正则化 不适定问题 |
Probability Density Estimation Based on Truncated Singular Value Decomposition Regularization |
| |
| Authors: | WU Di LIU Wen |
| |
| Affiliation: | Postgraduate;School of Science,WUT,Wuhan 430070,China |
| |
| Abstract: | The estimation of probability density is regarded as a linear operator equation from the definition of probability density.In this paper,truncated singular value decomposition（TSVD） regularization method of density estimation was proposed based on analyzing the singular value properties of kernel matrix of the operator equation.Compared with TSVD,the linearized Bregman iteration regularization method was introduced.The simulation indicates that the TSVD has better fitting results than the linearized Bregman iteration regularization method and has stronger robustness for noise. |
| |
| Keywords: | probability density estimation truncated singular value decomposition（TSVD） Bregman iteration regularization ill-posed problems |
| 本文献已被 维普 等数据库收录！ |
|
