SOURCE-LINKED INTELLIGENCE

Preprocessing Failure and Adversarial Detection in Depthwise-Separable Edge Vision Systems

arXiv · AI, language, vision and robotics · article · Sep 3, 2026 · UTC

Preprocessing-based defenses are the standard first-line response to adversarial attacks on edge vision systems, requiring no retraining, no architectural changes, and widely recommended as model-agnostic mitigations. Yet the foundational evaluations of these defenses were conducted on residual or Inception-class architectures, not on the depthwise-separable CNNs that dominate edge deployments. This untested assumption leaves a gap in the security evaluation literature. This paper closes that gap by evaluating six preprocessing defenses against adversarial perturbations across both architectur

Read original source ↗ Open in workspace

recordType

paper

region

Global