Blind separation of sources based on genetic algorithm
Yufang Yue, Jianqin Mao · 2003
Based on a genetic algorithm and variance normalization of output signals, this paper proposes two new methods for blind separation of sources. Simulation results illustrate that the method using a hybrid genetic algorithm not only keeps unsupervised, adaptive learning and the robust advantages of the known improved Herault-Jutten (H-J) (Jutten and Herault, 1991) algorithm, but also guarantees global convergence and less training time.