基于负熵的无预白化的独立成分分析算法

Independent Component Analysis Based on Negentropy Without Prewhitening

提出了一种基于负熵的无预白化的具有等变性的独立成分分析算法。首先介绍了负熵的定义以及如何使用负熵作为代价函数来度量混合信号的非高斯性。详细介绍了将分离向量和预白化矩阵结合得到了基于负熵的无预白化的迭代算法。最后通过选取4组非高斯向量作为信号源进行仿真实验,结果显示分离效果良好。

In this paper,a independent component analysis algorithm based on negentropy without prewhitening is present.First,the definition of negentropy is reviewed and how to use negentropy as a measure of cost function to mix the signal of nongaussian.Then,we propose the new fast-Ica algorithm based on negentropy without prewhitening,by combing the separating vector in fast-ica algorithm with the prewhitening matrix.Finally,the effectiveness of this new algorithm is verified by selecting 4 groups of non-gaussian vectors as signal sources.