Vision foundation models from OpenAI and Meta are non-robust to nine categories of common perturbations, with new metrics linking robustness scores to downstream performance drops and a fine-tuning method proposed to improve stability without losing utility.
Smallest enclosing disks (balls and ellipsoids)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Robustness of Vision Foundation Models to Common Perturbations
Vision foundation models from OpenAI and Meta are non-robust to nine categories of common perturbations, with new metrics linking robustness scores to downstream performance drops and a fine-tuning method proposed to improve stability without losing utility.