Part of master's degree.
Worked on analyisis of node classification techniques in graph machine learning. Tested on various datasets Node2Vec, GraphSAGE, GCN and GAT.
Here are some of the results:
-
GAT accuracy 82.1% on CORA dataset - visualized with tSNE method:
-
GCN accuracy of 76% on CORA dataset - visualized with tSNE method:
-
Node2Vec accuracy of 73% on CORA - visualized with tSNE method:
