Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[ZH] Fixing some issue on 13-2 #698

Merged
merged 1 commit into from
Dec 11, 2020
Merged

[ZH] Fixing some issue on 13-2 #698

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 3 additions & 5 deletions docs/zh/week13/13-2.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -115,9 +115,7 @@ $$
<!-- Now we have stability under coefficient perturbation. -->
現在我們就在系数扰动(coefficient perturbation)下有穩定性。

ChebNets are GCNs that can be used for any arbitrary graph domain, but the limitation is that they are isotropic. Standard ConvNets produce *anisotropic* filters because Euclidean grids have direction, while Spectral GCNs compute *isotropic* filters since graphs have no notion of direction (up, down, left, right).

<!-- We can extend ChebNets to multiple graphs using a 2D spectral filter. This may be useful, for example, in recommender systems where we have movie graphs and user graphs. Multi-graph ChebNets have the activation equation as below. -->
<!--ChebNets are GCNs that can be used for any arbitrary graph domain, but the limitation is that they are isotropic. Standard ConvNets produce *anisotropic* filters because Euclidean grids have direction, while Spectral GCNs compute *isotropic* filters since graphs have no notion of direction (up, down, left, right). We can extend ChebNets to multiple graphs using a 2D spectral filter. This may be useful, for example, in recommender systems where we have movie graphs and user graphs. Multi-graph ChebNets have the activation equation as below. -->
切布网路們都是圖形卷積網絡，它能夠用在任何任意的圖形領域上，但限制是它們都是各向同性的。標準的卷積网生成各向異性的過濾器，那是因為歐幾里得格子(Euclidean grids)是有方向的，同時光譜圖形卷積網絡計算各向同性過濾器，這是因為國形沒有方向這個概念，比如上下左右。我們可以延伸切佈網路到多重使用一個2d光譜過濾器的圖形。這或許是有用的，比如，在推薦系統中，我們有影片圖形和用戶圖形。多重圖形切佈網路有激活方程，下方所示。

$$
Expand Down Expand Up @@ -243,7 +241,7 @@ So, the activation of the next layer $h_{i}^{l+1}$ is a function of the activati
所以，下一层$h_{i}^{l+1}$的这个激活是...，就是在节点$i$和节点$i$的邻域的之前那一层$h_{i}^{l}$的激活的函数。当我们去改变这个函数时，我们就有整个图形的家族


### ChebNets and Vanilla Spatial GCNs
<!-- ### ChebNets and Vanilla Spatial GCNs -->
### 切布网络(ChebNets)和基础式空间性图形卷积网络(Vanilla Spatial GCNs)

<!-- The above defined Vanilla Spatial GCN is a simplification of ChebNets. We can truncate the expansion of ChebNet by using the first two Chebyshev functions to end up with, -->
Expand Down Expand Up @@ -297,7 +295,7 @@ $$h_{i}^{l+1} = (1 + \epsilon)h_{i}^{l} + \sum_{j \in N_{i}} h_{j}^{l}$$
<center>
<img src="{{site.baseurl}}/images/week13/13-2/Figure5.png" /><br>
<!-- <b>Figure 5:</b> Examples of two isomorphic graphs -->
<b>图5:</b> : 两个同构图形的例子
<b>图5:</b> 两个同构图形的例子


<!-- ## [Anisotropic GCNs](https://www.youtube.com/watch?v=Iiv9R6BjxHM&list=PLLHTzKZzVU9eaEyErdV26ikyolxOsz6mq&index=24&t=5586s) -->
Expand Down