摘要
基于广义Gauss概率密度模型对超Gauss和亚Gauss源混合信号进行盲分离,在Stiefel流形的自然梯度法框架下,分析了独立成分分析算法的局部稳定性,并采用几种典型非线性函数验证了所得结论的正确性和适用性.
Generalized Gauss density model was employed for the blind separation of the super-Gauss sources, and the local stability analysis was performed for independent algorithm in the framework of the natural gradient in Stiefel manifold. We presented some local stability analysis and verified their correctness and applicability. mixtures of component new results sub- and analysis about its
出处
《吉林大学学报(理学版)》
CAS
CSCD
北大核心
2012年第4期793-797,共5页
Journal of Jilin University:Science Edition
基金
国家自然科学基金(批准号:61071188
40776006)
湖北省自然科学基金(批准号:2009CDB077)
关键词
广义Gauss分布
独立成分分析
自然梯度
局部稳定性
generalized Gauss distribution
independent component analysis
natural gradient
local stability