Blind source separation of Fast ICA algorithm based on Negentropy and its optimization
Yi Wan, Yanghong Zhou, Zichen Yang · 2023
Blind Source Separation (BSS) is the problem of dividing a set of source signals from a mixed one. At present, Independent Component Analysis (ICA) is the main method to solve such problems. In this paper, the ICA algorithm based on maximum likelihood estimation is introduced at first. On the basis of summarizing its principle, we implement the data processing with MATLAB code. Then we give the FastICA algorithm based on Negentropy as an upgraded ICA. Further, by optimizing the distribution function, we derive the optimizing form of FastICA. The experimental simulation of this optimization algorithm is carried out with the previous two algorithms. After we analyze the experimental data, we confirm the superiority of the FastICA algorithm and the effectiveness of the optimized FastICA algorithm, and finally solve the cocktail party problem mentioned in the introduction.