Skip to content

Commit

Permalink
add conv p q desc
Browse files Browse the repository at this point in the history
  • Loading branch information
astonzhang committed Aug 26, 2018
1 parent c6f4ec1 commit 6ba3bdd
Show file tree
Hide file tree
Showing 2 changed files with 8 additions and 1 deletion.
6 changes: 5 additions & 1 deletion TERMINOLOGY.md
Original file line number Diff line number Diff line change
Expand Up @@ -72,6 +72,8 @@

归一化,normalization

过滤器,filter

过拟合,underfitting

假设,hypothesis
Expand Down Expand Up @@ -156,12 +158,14 @@ softmax回归,softmax regression

跳字模型,skip-gram model

调参,tune parameter
调参,tune hyper-parameter

通道,channel

凸优化,convex optimization

图像,image

未知词符号,unknown token

无偏估计,unbiased estimate
Expand Down
3 changes: 3 additions & 0 deletions chapter_convolutional-neural-networks/conv-layer.md
Original file line number Diff line number Diff line change
Expand Up @@ -64,6 +64,9 @@ class Conv2D(nn.Block):
return corr2d(x, self.weight.data()) + self.bias.data()
```

卷积窗口形状为$p \times q$的卷积层称为$p \times q$卷积层。同样,$p \times q$卷积或$p \times q$卷积核说明卷积核的高和宽分别为$p$和$q$。在有些文献中,卷积核也称过滤器(filter)。


## 图像中物体边缘检测

下面我们来看一个应用卷积层的简单应用:检测图像中物体的边缘,即找到像素变化的位置。首先我们构造一张$6\times 8$的图像(即高和宽分别为6和8像素的图像)。它中间4列为黑(0),其余为白(1)。
Expand Down

0 comments on commit 6ba3bdd

Please sign in to comment.