MorphStrata: Layer-Specific Perturbations for Generating Morphence Students in Time-Series Moving Target Defense
Researchers have developed MorphStrata, a novel defense strategy for time-series forecasting models against adversarial attacks. This method involves injecting selective, layer-specific stochastic noise into student models, creating structured heterogeneity. MorphStrata aims to enhance robustness without significantly increasing training overhead, showing promising results in maintaining adversarial RMSE across various datasets and attack scenarios. AI
IMPACT Introduces a novel defense mechanism that could improve the reliability of time-series forecasting models against adversarial attacks.