English(EN)The Fast Mixing Mechanism for Differential Privacy
新研究推动机器学习中的差分隐私发展
作者PulseAugur 编辑部·[5 个来源]·
研究人员开发了增强机器学习中差分隐私的新方法，特别是在去中心化和因果结构学习方面。一种名为DPDL的方法，使用基于相似性的校准技术和高斯噪声来保护非IID数据上的去中心化学习的隐私，实现了线性加速。另一种方法利用全同态加密（FHE）对加密数据进行因果结构学习计算，通过电路简化和近似来管理FHE的计算成本。此外，一种基于快速变换的新型DP草图机制为DP线性回归提供了更快的运行时间，而具有相关噪声的局部差分隐私模型表明，在不牺牲效用的情况下可以实现最优的集中式隐私成本。
AI
arXiv:2606.04399v1 Announce Type: new Abstract: In the paradigm of decentralized learning, a group of agents collaborate to train a global model using distributed datasets without a central server. Although the power of collaboration has been verified by many state-of-the-art stu…
arXiv:2606.05129v1 Announce Type: cross Abstract: Preserving data privacy is an important topic in structural data management and data mining. However, the issue of privacy leakage in distributed causal structure learning is a persistent challenge, especially in cases where data …
Preserving data privacy is an important topic in structural data management and data mining. However, the issue of privacy leakage in distributed causal structure learning is a persistent challenge, especially in cases where data transmission and computation are required. In this…
arXiv cs.LG
TIER_1English(EN)·Omri Lev, Moshe Shenfeld, Vishwak Srinivasan, Katrina Ligett, Ashia C. Wilson·
arXiv:2605.30600v1 Announce Type: new Abstract: Randomized sketching is a central tool for compressing large-scale optimization problems while preserving accuracy. In particular, sketches that are based on structured matrices, such as the Hadamard matrix, can be applied efficient…
arXiv cs.LG
TIER_1English(EN)·Madhura Pathegama, Srikanth Avasarala, Viveck R. Cadambe, Juba Ziani·
arXiv:2605.30476v1 Announce Type: cross Abstract: We study privately estimating the sum of $n$ user-held values in the presence of an honest-but-curious server. This motivates requiring privacy not only at data release but also throughout server-side computation. We therefore ado…
