往后其余的学习循环主要的 日地是细部地调整特征映像图,因此称之为算法的“收敛阶段”(Convergence Phase),收敛阶段时的学习率参数,应保持在相当小的数值(如0.01或更小), 而且收敛阶段通常需要大约数千次的学习循环。
基于20个网页-相关网页
训练阶段的相关算法与工作阶段的LMS算法可结合在同一个横向滤波器结构中,从而加速了自适应信道估计器的收敛过程。
The correlation algorithm in training period and the LMS algorithm in run period can be combined in the identical transversal filter structure, so that it speeds up the channel estimator convergence.
保证了失调误差较小,同时使算法在自适应初始阶段有较快的收敛速度。
It ensure the misadjustment is small and the algorithm has a high speed of convergence in the forepart of adaptation.
同时,两阶段的搜索形式使算法的收敛精度和搜索效率得到了保证。
Meanwhile, the two-stage search guarantees convergence accuracy and search efficiency of the algorithm.
应用推荐