bpnn icon indicating copy to clipboard operation
bpnn copied to clipboard

bp 神经网络算法

Results 2 bpnn issues
Sort by recently updated
recently updated
newest added

class Layer下的get_error函数,我认为_error内容应该是这样的: neru0=self.units[j] sum=0 for i in range(len(neru0.weight)): sum+=neru0.weight[i]*deltas[i] return sum 因为隐含层权值调整应该是同一个隐含层神经元对所有输出层神经元误差的求和,不是一个输出层神经元对所有隐含层的求和

Teach network XOR function pat = [ [[27.0], [30.0]], [[26.8], [30.0]], [[26.7], [30.1]], [[26.5], [30.1]], [[30.3], [31.9]], [[30.0], [30.0]], [[29.7], [30.0]], [[28.8], [31.7]], [[28.7], [31.7]], [[28.7], [31.5]] ] 1维数据该怎么填? 例如27.0->30.0