我打算为我的深度学习神经网络程序创建迷你批次，训练集包含’m’个样本。我尝试过以下方法：

# 首先打乱（X, Y）
permutation = list(np.random.permutation(m))
shuffled_X = X[:, permutation]
shuffled_Y = Y[:, permutation].reshape((1,m))
# 分割（shuffled_X, shuffled_Y）。注意最后一个迷你批次可能包含较少的训练样本。
num_complete_minibatches = math.floor(m/mini_batch_size) # 在你的分割中，迷你批次大小为mini_batch_size的数量
for k in range(0, num_complete_minibatches):
    ### START CODE HERE ### (约2行)
    mini_batch_X = shuffled_X[mini_batch_size*k:mini_batch_size*(k+2)]
    mini_batch_Y = shuffled_Y[mini_batch_size*k:mini_batch_size*(k+2)]

但这给我带来了以下结果：

shape of the 1st mini_batch_X: (128, 148)
shape of the 2nd mini_batch_X: (128, 148)
shape of the 3rd mini_batch_X: (12288, 148)
shape of the 1st mini_batch_Y: (1, 148)
shape of the 2nd mini_batch_Y: (0, 148)
shape of the 3rd mini_batch_Y: (1, 148)
mini batch sanity check: [ 0.90085595 -0.7612069   0.2344157 ]

预期输出是：

shape of the 1st mini_batch_X   (12288, 64)
shape of the 2nd mini_batch_X   (12288, 64)
shape of the 3rd mini_batch_X   (12288, 20)
shape of the 1st mini_batch_Y   (1, 64)
shape of the 2nd mini_batch_Y   (1, 64)
shape of the 3rd mini_batch_Y   (1, 20)
mini batch sanity check [ 0.90085595 -0.7612069 0.2344157 ] 

我确定我的切片实现有问题，但无法找出原因。任何帮助都非常感激。谢谢！

回答：

我认为您没有正确地切片numpy数组。最初打乱数组的方式是正确的。您不想切片第一维，所以保持它不变，使用:，并使用<Start Index>:<End Index>切片第二维。这是我在下面的代码中所做的。

for k in range(num_complete_minibatches+1):
### START CODE HERE ### (约2行)
    mini_batch_X = shuffled_X[:,mini_batch_size*(k):mini_batch_size*(k+1)]
    mini_batch_Y = shuffled_Y[:,mini_batch_size*(k):mini_batch_size*(k+1)]
    print(mini_batch_X.shape,mini_batch_Y.shape)