simpleGNN

I applied Graph Neural Network (GNN) on self-made dataset

I designed a simple, easy-to-understand and small dataset like zachary's karate club.

dataset description

We have 8 individuals, each follow in Instagram cause an edge in our graph. which you can add value to each friendship too.

• in feature matrix, first column shows number of posts, second one shows 'number of followers' and third one shows 'number of followings'

we want to predict if node 1 and node 8 split up, which nodes advocate either node 1 or node 8

References

• Blog GCNs by Thomas Kipf
• Semi-Supervised Classification with Graph Convolutional Networks by Thomas Kipf and Max Welling