Skip to main content
Account

Table 11 ERS ranking stability under random convex weights (illustrative sensitivity analysis)

From: Evaluating exposure-driven security governance for proactive risk mitigation of AI-powered cyber attacks in digital infrastructures

Illustrative setting

Weight sampling

Spearman ρ vs baseline (median [p10, p90])

Top 10 overlap vs baseline (mean/10)

Moderate component co-movement (synthetic exposures; n = 200)

w ∼ Dirichlet (1), N = 5000

0.92 [0.84, 0.98]

8.6/10

Stronger component co-movement (synthetic exposures; n = 200)

w ∼ Dirichlet (1), N = 5000

0.95 [0.88, 0.99]

9.0/10