Independent component analysis, a new concept?
Signal Processing - Special issue on higher order statistics
Independent component analysis: algorithms and applications
Neural Networks
Hi-index | 0.00 |
Independent component analysis (ICA) is an important topic of signal processing and neural network which transforms an observed multidimensional random vector into components that are mutually as independent as possible. In this paper, we have introduced a new method called SwiPe-ICA (S tepwi se Pe arsonian ICA) that combines the methodology of projection pursuit with Pearsonian density estimation. Pearsonian density function instead of the classical polynomial density expansions is employed to approximate the density along each one-dimensional projection using differential entropy. This approximation of entropy is more exact than the classical approximation based on the polynomial density expansions when the source signals are supergaussian. The validity of the new algorithm is verified by computer simulation.