摘要
给出一种快速感知器学习算法——变步长感知器学习算法 ,这种学习算法依公式 W(k+1) =W(k) -2 W(k) XTk X(k)调整权值和阈值。定理表明这种算法必然在有限步内收敛。
In this paper a new learning algorithm of perception with variable paces is given which adjusts its weights and threshold values according to the following formula: W(k+1)=W(k)-2W(k)XT kX(k). A theorem shows that the algorithm must be convergente after finite steps.
出处
《机械科学与技术》
CSCD
北大核心
2001年第2期234-235,241,共3页
Mechanical Science and Technology for Aerospace Engineering
基金
国家863计划项目! (863 -5 11-945 -0 11)
