| 非线性LMS算法实现盲源分离 |
| |
| 引用本文： | 倪晋平,陈亚林,马时亮. 非线性LMS算法实现盲源分离[J]. 西安工业大学学报, 2006, 26(5): 413-416 |
| |
| 作者姓名： | 倪晋平 陈亚林 马时亮 |
| |
| 作者单位： | 西安工业大学光电工程学院，西北工业大学，西安工业大学光电工程学院 西安710032，西安710032 |
| |
| 摘 要： | 在经典的最小均方(LMS)算法中引入非线性函数,得到一种非线性LMS算法.该算法根据LMS算法建立了相应的对照函数,用自然梯度推导出了自适应盲源分离算法,并且分别给出了实数算法和复数算法的详细推导过程.结果发现得到的算法即为最大化熵和最小化互信息(ACY)算法,从而揭示了ACY算法与LMS算法的关系,同时也得到ACY算法的复数形式.计算机仿真试验验证了算法的有效性.
|
| 关 键 词： | 盲源分离 LMS算法 非线性 随机梯度降 |
| 文章编号： | 1000-5714（2006）05-413-04 |
| 收稿时间： | 2006-05-10 |
| 修稿时间： | 2006-05-10 |
Non-linear LMS Algorithms Realize Blind Source Separation |
| |
| NI Jin-ping,CHEN Ya-lin,MA Shi-liang. Non-linear LMS Algorithms Realize Blind Source Separation[J]. Journal of Xi'an Institute of Technology, 2006, 26(5): 413-416 |
| |
| Authors: | NI Jin-ping CHEN Ya-lin MA Shi-liang |
| |
| Abstract: | The non-linear LMS algorithm is obtained by introducing the non-linearity into the typical LMS algorithm.The cost function is built according to LMS algorithm,the blind separation algorithm is induced,and the induction procession is given in case of real number and complex number respectively.The result indicates that the algorithm is the same to both maximum entropy and minimum mutual information(ACY) algorithms.The relation between ACY algorithms and LMS algorithms was proved.And the complex ACY algorithm was found out.The performance of proposed algorithms is verified by computer simulation whose results are shown in the paper. |
| |
| Keywords: | blind source separation LMS algorithm non-linear stochastic gradient descending |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
