中文说明：基于负熵和基于峭度的算法的相似系数和信干比相差不多，都能很好地实现信号的盲分离。基于峭度的FastICA算法的收敛速度要快，迭代次数比基于负熵的FastICA算法少四倍以上。从分离时间也可以看出基于峭度的算法比基于负熵的算法快将近四倍。SIR随信噪比增大的结果图可以看出存在高斯噪声时，基于峭度的FastICA算法的SIR略高于基于负熵的算法，最大相差1dB。SMSE随信噪比增大两种判据下的FastICA算法都逐渐变小，但是基于峭度的算法的SMSE更小，一般相差0.2左右。综上可知，基于峭度的FastICA算法性能要优于基于负熵的FastICA算法。
English Description：
Based on negative entropy and algorithms based on kurtosis coefficient of similarity and difference signal to interference ratio not much can realize blind signal separation. FastICA algorithm based on kurtosis faster convergence, iteration count four times less than the FastICA algorithm based on negative entropy times. Separation algorithm based on kurtosis can be seen nearly four times faster than the algorithm based on negative entropy. SIR with the result of increasing signal to noise ratio here you can see that when there is a Gaussian noise, SIR FastICA algorithm based on kurtosis above algorithm based on negative entropy, maximum difference of 1dB. Signal to noise ratio increases the SMSE FastICA algorithm under two criteria are getting smaller, but based on kurtosis SMSE smaller difference of 0.2 per cent. To sum up, FastICA algorithm based on kurtosis than the FastICA algorithm based on negative entropy.