JECA^2 is a new white-box attack method using Grad-CAM-guided perturbations and prompt embedding optimization to achieve judgment-explanation consistent adversarial attacks on forensic VLMs.
arXiv preprint arXiv:2408.13461 (2024)
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
VLA models exhibit joint-dependent success degradation under realistic physical faults, which J-PARC mitigates via latent regime inference and residual action correction.
citing papers explorer
-
Uncovering Vulnerability of Vision-Language-Action Models under Joint-Level Physical Faults
VLA models exhibit joint-dependent success degradation under realistic physical faults, which J-PARC mitigates via latent regime inference and residual action correction.