Analysis on The Stability of Minimum Mutual Information Algorithm
Chen Cui · Shipboard Electronic Countermeasure · 2010
The minimum mutual information algorithm is a typical algorithm of blind source separation,has better practicability.Through the simulation experiment,this paper shows:in order to assure the convergence of the algorithm,it need not to know the exact probability density function of the source signal,only need to satisfy the stability conditions.Further step,this paper analyzes and studies the conditions for the stability of the algorithm,and finds out that under different activation functions,the convergence speed of the algorithm is different although different activation functions can be selected.According to the difference between sup-Gaussian and sub-Gaussian characteristics of source signal,several common activation functions are summed up.Finally,the influences of Gaussian source and noise on the stability of minimum mutual information algorithm are discussed in this paper.