推广的感知器学习算法及其在模式识别中的应用

摘要

将感知器学习算法作了推广，将权向量和阀值的调整公式改为 W(k+1)=W(k)-αW(k)ZTkZk，其中α∈(1,2).证明了这种算法的收敛性，且可用于凸凹区域的识别中.%This paper generalized the learning algorithm of perception andadjusts weights and threshold according to the formulation W(k+1)=W(k)-αW(k)ZTkZk with α∈(1,2). It can be proved that the corresponding algorithm is convergent and can be used in the recognition of conve x and concave regions.