研究人员为正则化分类开发了新的采样界限，在广泛的 Lipschitz 连续损失函数上实现了最优的 $(1\pm\varepsilon)$-相对误差。该研究提出了改进的采样复杂度界限，具体来说，对于 L2 正则化是 $k^2/\varepsilon^2$，对于 L1 正则化是 $k/\varepsilon^2$。这些发现依赖于简单的均匀或范数采样，并且通过改进的论证来避免重复计数问题，显著优于之前的敏感性采样界限。 AI
影响 为分类算法中的采样效率建立了新的理论基准，可能影响未来机器学习系统的设计。
排序理由 详细介绍机器学习新理论结果的学术论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2605.23726v1 Announce Type: cross Abstract: We prove optimal sampling bounds achieving $(1\pm\varepsilon)$-relative error for a broad class of Lipschitz continuous classification loss functions under various regularization terms. This includes important functions such as lo…
We prove optimal sampling bounds achieving $(1\pm\varepsilon)$-relative error for a broad class of Lipschitz continuous classification loss functions under various regularization terms. This includes important functions such as logistic and sigmoid loss, hinge loss, and ReLU loss…