Master's Thesis - Neighborhood Sampling of Heterogeneous Graphs in Graph Neural Networks.
Paper: TODO
There are two parts that should be run separately due to differences in the version of the libraries (mainly PyTorch). The first one is heterogeneous random walks, and the second one is a graph neural network.
Heterogeneous random walks:
- PyTorch:
1.5.0a0+4ff3872, build from source
Graph neural network:
- PyTorch:
1.9.0