jrtechs
/
jrtechs-cs224w-notes
mirror of https://github.com/jrtechs/cs224w-notes.git
1
0
Fork 0
Code Issues 0 Releases 0 Wiki Activity
Browse Source

Update motifs-and-structral-roles_lecture.md

master
microLizzy 5 years ago
committed by GitHub
parent
ce829a4ee9
commit
46814a1a78
No known key found for this signature in database GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 2 additions and 1 deletions
Unified View
Diff Options
Show Stats Download Patch File Download Diff File
  1. +2
    -1
      preliminaries/motifs-and-structral-roles_lecture.md

+ 2
- 1
preliminaries/motifs-and-structral-roles_lecture.md View File

@ -59,11 +59,12 @@ We can consider roles as functions of nodes in a network which can be measured b
### Structural equivalence ### Structural equivalence
We say nodes u and v are structurally equivalent if they have the same relationships to each other. Structurally equivalent nodes are likely to be similar in many different ways. For example, node u and v are structurally equivalent in Figure 4 since they connect other nodes in the same way. We say nodes u and v are structurally equivalent if they have the same relationships to each other. Structurally equivalent nodes are likely to be similar in many different ways. For example, node u and v are structurally equivalent in Figure 4 since they connect other nodes in the same way.
![Figure 4](../assets/img/Exact_Subgraph_Enumeration.png?style=centerme)
![Figure 4](../assets/img/structurally_equivalent.png?style=centerme)
### RoIX ### RoIX
Roles allow us to identify different properties of nodes in network. Here we will introduce an automatic structural roles discovery method called RolX. It's an unsupervised learning approach without prior knowledge. Figure 5 is the RoIX approach overview. Roles allow us to identify different properties of nodes in network. Here we will introduce an automatic structural roles discovery method called RolX. It's an unsupervised learning approach without prior knowledge. Figure 5 is the RoIX approach overview.
![Figure 5](../assets/img/RoIX.png?style=centerme)
### Recursive Feature Extraction ### Recursive Feature Extraction
The basic idea of recursive feature extraction is to aggregate features of a node and use them to generate new recursive features. By this way we can turn network connectivity into structural features. The basic idea of recursive feature extraction is to aggregate features of a node and use them to generate new recursive features. By this way we can turn network connectivity into structural features.

Write Preview
Loading…
Cancel
Save