摘要
由于收敛速度快,快速独立分量分析(FastICA)被普遍运用,但为克服FastICA对初始值的选择较敏感的缺点,提出了将具有五阶收敛速度的牛顿迭代法和带有阻尼因子的牛顿迭代法引到算法中,形成两种方法有机结合的改进算法,并将改进的算法应用到仿真实验中,从而提高算法的稳定性.
The FastICA was widely used in because of its fast convergence property.In order to overcome its shortcoming of sensitivity to initial value problem,the Newton iterative method with five order convergence speed and with a damping factor of the Newton iterative method was proposed.The two methods were combined to form the final improvement algorithm,so the performance of the algorithm is improved.
出处
《佳木斯大学学报（自然科学版）》
CAS
2013年第4期579-581,585,共4页
Journal of Jiamusi University：Natural Science Edition
