摘要
扩展Infomax算法是在传统Infomax算法基础上发展起来的一种较为实用的盲源分离算法。该算法采用非线性模型的动态切换技术,实现了对超高斯源和亚高斯源的同步分离。该文结合模型切换矩阵系数和迭代误差在盲源分离过程中的变化,直观地展示了扩展Infomax算法的收敛过程。文中还探讨了高斯随机噪声和随机脉冲干扰对算法收敛性能的影响。所得结论对于扩展Infomax算法的实际应用有一定的指导意义。
Extended-Infomax algorithm is practical algorithm of blind source separation,which is developed from the traditional Infomax algorithm proposed by Bell in1995.The extended -Infomax algorithm,using the nonlinear model switching technique,can synchronously separate the super-and sub-Gaussian sources.Combining the dynamic changing of switching matrix and iterative errors,this paper intuitively displays the convergence procedure of extended -Infomax algorithm.It also discusses the influences of Gaussian source and random pulse on the convergence of extended-Infomax algorithm.The research results have directive meaning for practical use of extended-Infomax algorithm.
出处
《计算机工程与应用》
CSCD
北大核心
2003年第7期49-51,59,共4页
Computer Engineering and Applications
基金
国家自然科学基金(编号:60271024)
安徽省自然科学基金项目(编号:0043214)
