Research map / Federated learning and privacy / Differential privacy
Improved Bounds for Private and Robust Alignment
Wenqian Weng, Yi He, Xingyu Zhou
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use MLE-type log loss instead of debiased log loss for private alignment | private DPO-style alignment · finance_preference (synthetic, eps=0.5) | win rate vs SFT (%): 65.8 -> 65.2 | No clear effect |
| use MLE-type log loss instead of square loss for private alignment | private DPO-style alignment · finance_preference (synthetic, eps=0.5) | win rate vs SFT (%): 64.5 -> 65.2 | No clear effect |
| use MLE-type log loss instead of debiased log loss | private alignment with chiPO · PKU-SafeRLHF (eps=0.5) | win rate vs SFT (%): 61.95 -> 61.45 | No clear effect |
| use MLE-type log loss instead of square loss | private alignment with chiPO · PKU-SafeRLHF (eps=0.5) | win rate vs SFT (%): 61.64 -> 61.45 | No clear effect |
Closest papers by meaning
- Square$\chi$PO: Differentially Private and Robust $\chi^2$-Preference Optimization in Offline Direct Alignment (ICML 2025 · similarity 0.93)
- Oracle-Robust Online Alignment for Large Language Models (NeurIPS 2026 · similarity 0.89)
- Differentially Private Steering for Large Language Model Alignment (ICLR 2025 · similarity 0.89)
- A Unified Theoretical Analysis of Private and Robust Offline Alignment: from RLHF to DPO (ICML 2025 · similarity 0.88)
- Distributionally Robust Listwise Preference Optimization (NeurIPS 2026 · similarity 0.88)
- Mission Impossible: A Statistical Perspective on Jailbreaking LLMs (NeurIPS 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
