代码
Weights1 = tf.Variable(tf.random_normal([11, 4]))
biases1 = tf.Variable(tf.zeros([1, 4]) + 0.1)
Wx_plus_b1 = np.array(tf.matmul(l0, Weights1) + biases1)
N1act = 2/(1+pow(math.e,-Wx_plus_b1[3]))-1
prediction = tf_spiky(N1act)
输出层有4个输出,我要用第三个输出的值作为N1act的输入,感觉没错呀,请问该如何解决?