Hi everyone,
I’m just starting to learn about GNN samplers, but I’m struggling to fully understand what as_edge_prediction_sampler does.
Also, I’m curious if this sampler can be used with other sampling algorithms like VR-GCN or subgraph-based methods such as GraphSAINT. My goal is to classify the types of relationships that exist between the nodes in a graph.
In graph machine learning tasks like node classification or link prediction, message passing is performed on subgraphs sampled from the original large graph. as_edge_prediction_sampler constructs an edge-wise sampler based on the provided node-wise sampler. While the node-wise sampler sample a subgraph given a specific node, the edge-wise sampler applies corresponding logic to both the source node and the target node of a specific edge for subgraph sampling. There are some examples here to help understand it.
Since DGL 2.0, we recommend to use graphbolt for data loading, which should be easier to use and understand.
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.