研究人员发现差分隐私选择性更新和发布（DPSUR）算法的隐私核算存在缺陷。现有方法忽略了其选择性发布机制引入的采样概率变化，可能削弱隐私保证。为解决此问题，提出了一种名为基于梯度裁剪的差分隐私选择性发布（DPSR-CG）的新算法，该算法提供了更严格的隐私分析，并在各种数据集上表现出强劲的性能。 AI
影响 增强了在敏感数据上训练的机器学习模型的隐私保证，可能使其在受监管行业中得到更广泛的应用。
排序理由 该集群包含一篇详细介绍新算法及其分析的学术论文。
在 Hugging Face Daily Papers 阅读 →
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
Machine learning's reliance on sensitive data necessitates privacy-preserving techniques like Differentially Private Stochastic Gradient Descent (DPSGD). However, DPSGD suffers from substantial utility degradation and slow convergence due to gradient clipping and noise injection.…
arXiv:2606.04384v1 Announce Type: cross Abstract: Machine learning's reliance on sensitive data necessitates privacy-preserving techniques like Differentially Private Stochastic Gradient Descent (DPSGD). However, DPSGD suffers from substantial utility degradation and slow converg…
Machine learning's reliance on sensitive data necessitates privacy-preserving techniques like Differentially Private Stochastic Gradient Descent (DPSGD). However, DPSGD suffers from substantial utility degradation and slow convergence due to gradient clipping and noise injection.…