Conjugate Descent Algorithm of Independent Component Analysis
Luo Si · 2003
The Independent Component Analysis as a method widely used in blind source separation is a hotspot in signal processing.Based on analysis of existing algorithms, we give conjugate gradient algorithm using minimal mutual information criteria. We also compare the algorithm with the traditional algorithms in speed, which reduces times of iterations from about 2 000 to not more than 300, and analyse briefly criterion of selecting contrast function to realize global convergence. At last, the simulation experiments prove its good convergence and robust.