Degree-Preserving Rewiring
A graph perturbation technique that randomly swaps edge endpoints while keeping the degree of every node unchanged, producing control graphs with identical degree distributions but randomized adjacency, used to test whether graph structure is effectively utilized.
Timeline (last 90 days)
The diagnostic protocol uses degree-preserving rewiring to test whether neighbor identity truly matters.
Degree-preserving rewiring is a graph perturbation technique: while keeping each node's degree unchanged, it randomly swaps edge endpoints to generate a control graph with the same degree distribution as the original but randomized adjacency.
On the feeder data, adding edges to the same backbone yields a smaller gain, and more than half of it disappears after degree-preserving rewiring.
Before deploying a GNN for load forecasting, at least two references should be run: a per-node independent network and a shared-parameter network, and only when edges still bring stable gains under reconnection tests is there sufficient reason to maintain the graph structure.
All Facts (4)
The diagnostic protocol uses degree-preserving rewiring to test whether neighbor identity truly matters.
50%UnverifiedDegree-preserving rewiring is a graph perturbation technique: while keeping each node's degree unchanged, it randomly swaps edge endpoints to generate a control graph with the same degree distribution as the original but randomized adjacency.
50%UnverifiedOn the feeder data, adding edges to the same backbone yields a smaller gain, and more than half of it disappears after degree-preserving rewiring.
50%UnverifiedBefore deploying a GNN for load forecasting, at least two references should be run: a per-node independent network and a shared-parameter network, and only when edges still bring stable gains under reconnection tests is there sufficient reason to maintain the graph structure.
50%