You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Feb 27, 2021. It is now read-only.
举个例子,若做完了平均值池化,然后对(前一层的)池化(神经)单元进行简单的均匀分布(译者注:均匀分布,即 $P(X=k)=\frac{1}{m},k=1,...,m$ )上采样。在最大值池化中,即使(上层)输入(译者注:在池化的相邻区域内)的变化很小,(池化)单元会对结果产生扰动(译者注:本句翻译不确定,"In max pooling the unit which was chosen as the max receives all the error since very small changes in input would perturb the result only through that unit. ")。
Finally, to calculate the gradient w.r.t to the filter maps, we rely on the border handling convolution operation again and flip the error matrix δ(l)k the same way we flip the filters in the convolutional layer.