Neural Antidote: Class-Wise Prompt Tuning for Purifying Backdoors in CLIP

Documenti analoghi

• BadCLIP: Trigger-Aware Prompt Learning for Backdoor Attacks on CLIP
  di: Bai, Jiawang, et al.
  Pubblicazione: (2023)
• Revisiting Backdoor Attacks on LLMs: A Stealthy and Practical Poisoning Framework via Harmless Inputs
  di: Kong, Jiawei, et al.
  Pubblicazione: (2025)
• Retrievals Can Be Detrimental: Unveiling the Backdoor Vulnerability of Retrieval-Augmented Diffusion Models
  di: Fang, Hao, et al.
  Pubblicazione: (2025)
• Grounding Language with Vision: A Conditional Mutual Information Calibrated Decoding Strategy for Reducing Hallucinations in LVLMs
  di: Fang, Hao, et al.
  Pubblicazione: (2025)
• CLIP-Guided Generative Networks for Transferable Targeted Adversarial Attacks
  di: Fang, Hao, et al.
  Pubblicazione: (2024)