mini batch，cross entorpy error

import numpy as np

# cross entropy error mini batch, if t is one hot
def mini_batch_cee(y, t):
    batch_size = y.shape[0]
    return -np.sum(t * np.log(y + 1e-7))/batch_size


# cross entropy error mini batch, if t is not one hot, is lable
def mini_batch_cee_sec(y, t):
    if y.ndim == 1:  # 1维转变为多维
        t = t.reshape(1, t.size)
        y = y.reshape(1, y.size)

    batch_size = y.shape[0]
    # np.arange(batch_size) generate [0,1,2,...]
    # y[np.arange(batch_size), t] generate [y[0,0],y[0,1]...]
    return -np.sum(np.log(y[np.arange(batch_size), t] + 1e-7))/batch_size


# mini batch 
def mini_batch(y, t):
    y_size = y.shape[0]
    batch_size = 3
    batch_mask = np.random.choice(y_size, batch_size)  # 返回下标
    y_batch = y[batch_mask]
    t_batch = t[batch_mask]
    return y_batch, t_batch


y = np.array([0.1, 0.2, 0.6, 0.1])  # 模拟神经网络输出
t = np.array([0, 0, 1, 0])  # one-hot表示
t_1 = np.array([0, 1, 2, 3])  # lable

y_batch, t_batch = mini_batch(y, t)
print("batch select", y_batch, t_batch)

mini_batch_res = mini_batch_cee(y, t)
print("mini batch one hot", mini_batch_res)


mini_batch_res = mini_batch_cee_sec(y, t_1)
print("mini batch lable", mini_batch_res)