Beyond Gradient and Priors in Privacy Attacks: Leveraging Pooler Layer Inputs of Language Models in Federated Learning

Documenti analoghi

• DP-MemArc: Differential Privacy Transfer Learning for Memory Efficient Language Models
  di: Liu, Yanming, et al.
  Pubblicazione: (2024)
• Privacy-Preserving Data Deduplication for Enhancing Federated Learning of Language Models (Extended Version)
  di: Abadi, Aydin, et al.
  Pubblicazione: (2024)
• Geminio: Language-Guided Gradient Inversion Attacks in Federated Learning
  di: Shan, Junjie, et al.
  Pubblicazione: (2024)
• Continuous Embedding Attacks via Clipped Inputs in Jailbreaking Large Language Models
  di: Xu, Zihao, et al.
  Pubblicazione: (2024)
• Gradient Cuff: Detecting Jailbreak Attacks on Large Language Models by Exploring Refusal Loss Landscapes
  di: Hu, Xiaomeng, et al.
  Pubblicazione: (2024)