Adaptive Blind Signal Processing Based on Non-negative Matriz Factorization
In this paper, non-negative matrix factorization (NMF) technique is applied to adaptive blind signal separation, and the objective function with restricted conditions is proposed. Fist of all, natural gradient method is used in independent component analysis (ICA) of adaptive blind signal, and we get the robust estimation function of ICA and super-effectiveness of solution convolution, and then NMF algorithm is used to have related statistical blind independent source separation and we get second-order statistic with robust and blind identification of blind source matrix where the sources are independent each other, Finally we make use of state-space to give linear blind separation and get blind correlated solution, and we analyze how to extract the blind signal processing is analyzed by non-parametric power spectrum estimation and we examine the feasibility and effectiveness of blind source separation including statistical correlated and generalized Gaussian distribution signal source, and adaptive blind signal separation is better achieved.
non-negative matriz factorization blind signal separation independent component analysis adaptive generalized Gaussian distribution Introduction
Gang Liu Bo-ping Tian Feng Shan
School of science, Shengyang Institute of Aeronautical Engineering, P.R.China, 110136 Department of Mathematics, Harbin Institute of Technology, P.R.China, 150001
国际会议
The Third International Workshop on Applied Matriz Theory(第三届国际矩阵分析与应用会议)
杭州
英文
764-767
2009-07-09（万方平台首次上网日期，不代表论文的发表时间）