Random Edge Baseline
An ablation method in graph machine learning that keeps node features and model hyperparameters unchanged while randomly rewiring the edges of a graph before training and evaluation. It is used to separate the gain contributed by graph structural information from that contributed by additional features.
Timeline (last 90 days)
The random-edge baseline (keeping node features and model hyperparameters unchanged, only randomly rewiring edges) performs better than the basic neural network but clearly worse than the GNN that retains the true connection relationships.
The random-edge baseline is an ablation experiment used to separate the information of the graph structure itself from the gains brought by introducing additional features or more parameters.
All Facts (2)
The random-edge baseline (keeping node features and model hyperparameters unchanged, only randomly rewiring edges) performs better than the basic neural network but clearly worse than the GNN that retains the true connection relationships.
50%UnverifiedThe random-edge baseline is an ablation experiment used to separate the information of the graph structure itself from the gains brought by introducing additional features or more parameters.
50%