الملخص
Token-level differentially private (DP) prompt optimization methods such as DP-OPT can become unstable under tight privacy budgets: on GSM8K, DP-OPT obtains $49.5\pm28.5\%$ across 30 runs, and a logged search trajectory reveals prompt-template drift and noise-sensitive irreversible choices. We diagnose these as structural consequences of greedy token-by-token construction over privately aggregated counts. We then propose DP-ES (Differentially Private Evolution Strategies), a structurally cleaner alternative that maintains a population of full prompts, mutates them via LLM calls that never access the private dataset, and spends privacy only on sampled-Gaussian evaluation; deterministic or Gumbel-smoothed selection is post-processing. Under a conservative $(\varepsilon\leq1.0,δ=10^{-5})$ guarantee, DP-ES achieves 88.1% on GSM8K (+38.6 pp over DP-OPT, approximately 9 times lower standard deviation), 99.7% on MedQA, 73.5% on BANKING77, and 86.8% on Alpaca. It is also 2.5 times faster in wall-clock time and uses 3.3 times fewer logged private-data call groups than DP-OPT. Selection and population ablations, implementation-level noise checks, and a 200-profile exact-match memorization stress test complement the formal guarantee. Scope: Our experiments establish optimization robustness under DP noise, especially where prompt structure is critical; end-to-end validation on genuinely sensitive, non-saturated deployment data remains future work.
الكلمات المفتاحية
الموضوع
بيانات النشر
- المجلة
- غير متاح
- وصول مفتوح
- وصول مفتوح أخضر
اقتبس هذه المقالة
APA 7
Liu, Z., Li, A., Han, Y., Yu, H., Zhang, J., Zhu, D., Li, C., & Zhang, S. (2026). DP-ES: Differentially Private Evolution Strategies for Prompt Optimization. https://omanscience.com/ar/articles/dp-es-differentially-private-evolution-strategies-for-prompt-optimization
MLA 9
Liu, Ziniu, et al. "DP-ES: Differentially Private Evolution Strategies for Prompt Optimization." https://omanscience.com/ar/articles/dp-es-differentially-private-evolution-strategies-for-prompt-optimization.
شيكاغو (المؤلف–التاريخ)
Liu, Ziniu, Aiping Li, Yue Han, Han Yu, Junjian Zhang, Dong Zhu, Changjian Li, and Shiqiang Zhang. 2026. "DP-ES: Differentially Private Evolution Strategies for Prompt Optimization." https://omanscience.com/ar/articles/dp-es-differentially-private-evolution-strategies-for-prompt-optimization.
هارفارد
Liu, Z., Li, A., Han, Y., Yu, H., Zhang, J., Zhu, D., Li, C. and Zhang, S. (2026) 'DP-ES: Differentially Private Evolution Strategies for Prompt Optimization', Available at: https://omanscience.com/ar/articles/dp-es-differentially-private-evolution-strategies-for-prompt-optimization.
فانكوفر
Liu Z, Li A, Han Y, Yu H, Zhang J, Zhu D, et al. DP-ES: Differentially Private Evolution Strategies for Prompt Optimization. https://omanscience.com/ar/articles/dp-es-differentially-private-evolution-strategies-for-prompt-optimization
IEEE
Z. Liu, A. Li, Y. Han, H. Yu, J. Zhang, D. Zhu, C. Li, and S. Zhang, "DP-ES: Differentially Private Evolution Strategies for Prompt Optimization," https://omanscience.com/ar/articles/dp-es-differentially-private-evolution-strategies-for-prompt-optimization.