Statistical Inference for Independent Component Analysis Based on Polynomial Spline Model

This paper develops the confidence interval for the independent component analysis. The method is based on the bootstrap method using source density functions estimated by the polynomial splines modeling. A simulation study is conducted to show the numerical example for the proposed method and that the confidence interval has a reasonable coverage probability. Finally, the method is applied to a real fetal electrocardiogram data. One characteristic signal was effectively detected as a favor of the blind source separation by the proposed method.