近期论文6
- Understanding and Preserving Safety in Fine-Tuned LLMs第 4 位,共 10 位作者CCS 2026 · 7 次引用
- Locket: Robust Feature-Locking Technique for Language Models第 1 位,共 3 位作者ACL 2026 · 1 次引用
- SoK: Colluding Adversaries in Machine Learning Pipelines第 2 位,共 4 位作者USENIX Security 2026
- LookAhead: Preventing DeFi Attacks via Unveiling Adversarial Contracts第 2 位,共 7 位作者FSE 2025 · 3 次引用
- Activation Approximations Can Incur Safety Vulnerabilities in Aligned LLMs: Comprehensive Analysis and Defense第 3 位,共 10 位作者USENIX Security 2025
- Secure Transformer Inference Made Non-interactive第 3 位,共 10 位作者NDSS 2025
