Robustness of $\ell _1$-Norm Estimation: From Folklore to Fact

Counteracting adversarial attacks in autonomous driving

“…The pseudo-code of the overall optimization training flow is shown in Algorithm 1. The advantages of using ℓ 1 -norm over ℓ 2 -norm in terms of robustness analysis procedures are largely recognized across the scientific literature [37]. To improve model Compute L ∇ with δ l and δ r , according to Equation (15);…”