Fig 1.
Overall research workflow of this paper.
Fig 2.
An example of the cascading failure propagation process.
Fig 3.
Schematic of the cascading failure process.
Fig 4.
Flowchart of critical nodes identification process.
Table 1.
Statistical summary of USAir97.
Fig 5.
Basic topological information with centrality measures.
Table 2.
Total variance explained.
Table 3.
The performance evaluation results: Cohen’s d, Inference time and Spearman correlation.
Fig 6.
Performance evaluation on critical node identification.
Fig 7.
Changes in evaluation metrics of the GraphSAGE model under β -ablation experiments.
Fig 8.
Targeted vs. Random Attacks: GraphSAGE versus baseline methods.
Fig 9.
Comparing GraphSAGE to baseline methods for node ranking.
Table 4.
Changes in evaluation metrics of the GraphSAGE model: comparison of input feature vectors before and after PCA ablation.
Fig 10.
Sensitivity analysis of resilience metric weights.
Fig 11.
Effect of β on cascade failure ratio and network resilience.
Fig 12.
Targeted β tuning for improved resilience at critical nodes.