https://blog.csdn.net/weixin_43114885/article/details/90378914
4.2.4 mini-batch 版交叉熵误差的实现
one-hot格式的监督数据
def cross_entropy_error(y, t):
if y.ndim == 1: #一维数组要把第一维放到第二维,表示只有一条数据
t = t.reshape(1, t.size) #reshape函数代两维的参数
y = y.reshape(1, y.size)
batch_size = y.shape[0] #记下第一维的数值,表示有多少条数据
return -np.sum(t * np.log(y + 1e-7)) / batch_size
t为(batch_size,10), y 为(batch_size,10)
t*logy 结果还是(batch_size,10)
np.sum 后成为一个数值
y=np.array([[0,1,0.1,0,0,0,0,0,0,0],
[0,0,0.2,0.8,0,0,0,0,0,0]])
t=np.array([[0,1,0,0,0,0,0,0,0,0],
[0,0,0,1,0,0,0,0,0,0