DDSA: Dual-Domain Strategic Attack for Spatial-Temporal Efficiency in Adversarial Robustness Testing

TL;DR

DDSA tackles the scalability gap in adversarial robustness testing for real-time image streams by introducing a dual-domain approach that combines a scenario-aware temporal trigger with explainable AI–guided spatial targeting. By evaluating frames with a trigger score $Q_i$ and restricting perturbations to critical regions identified via Integrated Gradients, DDSA achieves substantial resource savings while maintaining high attack effectiveness on priority classes. Empirical results show 80–97% reductions in computation and robust attack performance even when perturbing as little as 30% of pixels, underscoring practical viability for large-scale, safety-critical deployments. This framework enables real-time, coverage-focused robustness testing in resource-constrained environments such as SAR, agricultural monitoring, and large-scale image classification pipelines.

Abstract

Image transmission and processing systems in resource-critical applications face significant challenges from adversarial perturbations that compromise mission-specific object classification. Current robustness testing methods require excessive computational resources through exhaustive frame-by-frame processing and full-image perturbations, proving impractical for large-scale deployments where massive image streams demand immediate processing. This paper presents DDSA (Dual-Domain Strategic Attack), a resource-efficient adversarial robustness testing framework that optimizes testing through temporal selectivity and spatial precision. We introduce a scenario-aware trigger function that identifies critical frames requiring robustness evaluation based on class priority and model uncertainty, and employ explainable AI techniques to locate influential pixel regions for targeted perturbation. Our dual-domain approach achieves substantial temporal-spatial resource conservation while maintaining attack effectiveness. The framework enables practical deployment of comprehensive adversarial robustness testing in resource-constrained real-time applications where computational efficiency directly impacts mission success.

Figures (5)

• Figure 1: Task-dependent classification priority in image signal processing systems.
• Figure 2: XAI-guided spatial targeting demonstration.
• Figure 3: Classifier loss under XAI-guided vs. random selection.
• Figure 4: Testing perturbation comparison: full-image vs. DDSA.
• Figure 5: DDSA achieves 80-97% computational time reduction.