Fragile by Design: On the Limits of Adversarial Defenses in Personalized Generation

Documenti analoghi

• TAIJI: Textual Anchoring for Immunizing Jailbreak Images in Vision Language Models
  di: Yin, Xiangyu, et al.
  Pubblicazione: (2025)
• Boosting Adversarial Training via Fisher-Rao Norm-based Regularization
  di: Yin, Xiangyu, et al.
  Pubblicazione: (2024)
• CeTAD: Towards Certified Toxicity-Aware Distance in Vision Language Models
  di: Yin, Xiangyu, et al.
  Pubblicazione: (2025)
• Towards Fairness-Aware Adversarial Learning
  di: Zhang, Yanghao, et al.
  Pubblicazione: (2024)
• A Black-Box Evaluation Framework for Semantic Robustness in Bird's Eye View Detection
  di: Wang, Fu, et al.
  Pubblicazione: (2024)