反向传播算法也称BP 算法。由于这种算法在本质上是一种神经网络学习的数学模型，所以，有时也称为BP 模型。 BP 算法是为了解决多层前向神经网络的权系数优化而提出来的；所以，BP 算法也通常暗示着神经网络的拓扑结构是一种无反馈的多层前向网络，它含有输人层、输出层以及处于输入输出层之间的中间层。中间层有单层或多层，由于它们和外界没有直接的联系，故也称为隐层。在隐层中的神经元也称隐单元。隐层虽然和外界不连接．但是，它们的状态则影响输入输出之间的关系。这也是说，改变隐层的权系数，可以改变整个多层神经网络的性能。 ' 程序实现功能：模拟QQ 截屏Private mW1() As Double ' 隐含层的权值S1 X R Private mW2() As Double ' 输出层的权值S2 X R Private mB1() As Double '隐含层的偏置值S1 X 1 Private mB2() As Double '输出层的偏置值S2 X 1 Private mErr() As Double '均方误差Private mMinMax() As Double '输入向量的上下限R X 2 Private mS1 As Long '隐含层的神经元个数S1 Private mS2 As Long '输出层的神经元个数S2 Private mR As Long '输入层神经元个数R Private mGoal As Double '收敛的精度Private mLr As Double '学习速度Private mGama As Double '动量系数Private mMaxEpochs As Long '最大的迭代次数Private mIteration As Long '实际的迭代次数'**************************************** 中间变量******************************************* Private HiddenOutput() As Double '隐含层的输出Private OutOutput() As Double '输出层的输出Private HiddenErr() As Double '隐含层各神经元的误差Private OutPutErr() As Double '输出层各神经元的误差Private Pdealing() As Double '当前正在处理的输入Private Tdealing() As Double '当前正在处理的输入对应的输出Private OldW1() As Double ' 旧权值数组Private OldW2() As Double ' 旧权值数组Private OldB1() As Double '旧偏置值数组Private OldB2() As Double '旧偏置值数组Private Ts As Long '输入向量的总个数Private Initialized As Boolean '是否已初始化'**************************************** 属性******************************************* Public Event Update(iteration) Public Prope...
