基于神经网络求解线性方程组的方法

Study on the Method of Solving Linear Equations Based on Neural Network

以单层神经网络为基础，利用δ规则进行训练，提出了一种求具有唯一解的线性方程组解的一种新方法，给出了δ学习规则的数学推导和求解线性方程组的ＳＮＮ（Ｓｉｎｇｌｅ－ｌａｙｅｒＮｅｕｒａｌＮｅｔｗｏｒｋ）学习算法，并利用该方法求解了大量线性方程组，实验结果表明该方法是有效的和可行的。

Based on the training of single-layer neural network and the rule of δ,A new method of solving linear equations is proposed. Meanwhile, the mathematical derivation of the rule of δ and the studying algorithm of SNN is present. Moreover, by use of this algorithm a lot of linear equations was solved, the results prove that the studying algorithm of SNN is effective and resonable to solve linear equations.