摘要
近年来,ICA(Independent Component Analysis,独立成分分析)已成为处理BSS(Blind Source Separation,盲源分离)问题的主要手段,同时也受到人们越来越多的关注,为此讨论ICA的原理及其优越性。首先介绍ICA,然后引入FastICA算法的推导过程,最后通过MATLAB仿真将其与梯度算法、PCA(Principal Component Analysis,主成分分析)算法所得的仿真结果进行对比分析。通过算法验证,经FastICA处理得到的分离信号与源信号相关系数的绝对值不小于0.99,与其他两种算法比较可以明显地得到FastICA是一种更为有效的盲源分离方法。
ICA has been a primary method solving BSS in recent years, and aroused more and more concern, so discuss the principle and superiority. In this paper,introduce ICA and FastICA algorithm firstly,then analyze simulation result by FastICA, gradient algorithm and PCA. Through verification,absolute value of correlation coefficient between separation signals and source signals is not less than 0.99. Compared with other algorithms,conclude FastICA is a more effective algorithm.
出处
《计算机技术与发展》
2011年第12期93-96,共4页
Computer Technology and Development
基金
国家自然科学基金(61032001
60972159
61002006)
关键词
独立成分分析
盲源分离
主成分分析
梯度算法
independent component analysis
Mind source separation
principal component analysis
gradient algorithm