摘要
从一般前馈神经网络模型出发，构造出一组关于权重的非线性方程组，给出不同于传统ＢＰ算法的新型神经元算法．理论证明了该算法的收敛性，从而避免了ＢＰ算法的局限性．
The paper constitutes a series of non-linear equations about weight based on general feedforward neural network model. A new algorithm different from traditional BP algorithm is proposed, whose convergence is proven, thus avoid the limitations of traditional BP algorithm.
出处
《控制与决策》
EI
CSCD
北大核心
1995年第3期284-288,共5页
Control and Decision
