IMPROVED FAST-ICA ALGORITHM BASED ON CONVERGENCE PROOF OF NONLINEAR FUNCTION

The selection of nonlinear function has a significant influence on the separation performance of Fast-ICA. This paper proves the convergence performance of the sine function as a Fast-ICA nonlinear function. The simulation results show that the proposed algorithm is suitable for an image signal, communication signal, sound signal, and linear frequency modulated signal. Furthermore, the performance of the algorithm according to the increase of the number of sources is analyzed.

Automated summary

The selection of the nonlinear function in Fast-ICA has a significant impact on the separation performance. This paper specifically demonstrates the convergence performance of the sine function in Fast-ICA and shows that the algorithm is suitable for various types of signals. Additionally, the algorithm's performance is analyzed as the number of sources increases.