约束独立成分分析模型及其算法综述

Overview of Model and Algorithms of Constrained Independent Component Analysis

约束独立成分分析(Constrained Independent Component Analysis,CICA)是在独立成分分析(Independent Component Analysis,ICA)基础上,发展起来的一类信号分离(提取)技术。CICA充分利用源信号的先验信息,从而有效地解决ICA应用中出现的不确定性、计算量大和内存空间浪费等问题。首先,简述了CICA的基本模型以及源信号较普遍的先验信息:非负性、稀疏性和时间结构;然后,重点介绍了基于每一种先验信息的CICA算法,并总结了CICA模型的应用;最后,展望CICA模型的未来改进方向。

Constrained independent component analysis（CICA） is a source separation and extraction method,which is based on the independent component analysis（ICA）. Making full use of the prior information about the source, CICA can be more efficient to resolve the problems emerged in the ICA applications, such as the uncertainty, large calculation, the waste of memory space, and so forth. Firstly, the basic model of CICA and the universal prior information of source, namely nonnegativity, sparsity and temporal structure, are described. Then CICA algorithm is analyzed in detail corresponding each prior information, and the applications of CICA are summarized. Finally, some remarks on the future research of CICA are presented.