PulseAugur
实时 09:37:28

Linear-Core Surrogates 提供用于分类的平滑损失函数和线性速率

研究人员推出了一种新型凸损失函数家族——Linear-Core (LC) Surrogates,旨在结合机器学习中平滑损失和分段线性损失的优点。这些代理是可微的,并实现了线性一致性界限,提供了更高的统计效率。在结构化预测任务中,LC Surrogates 能够实现更高效的随机梯度估计器,避免了二次复杂度,从而节省了大量的计算和能源。 AI

影响 引入了一种新的损失函数家族,提高了优化速度和统计效率,有可能加速结构化预测任务的训练并降低能耗。

排序理由 学术论文,介绍了一类具有理论和实践优势的新型损失函数。

在 arXiv stat.ML 阅读 →

AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →

Linear-Core Surrogates 提供用于分类的平滑损失函数和线性速率

报道来源 [2]

  1. arXiv stat.ML TIER_1 English(EN) · Mehryar Mohri, Yutao Zhong ·

    Linear-Core Surrogates: Smooth Loss Functions with Linear Rates for Classification and Structured Prediction

    arXiv:2604.27742v1 Announce Type: cross Abstract: The choice of loss function in classification involves a fundamental trade-off: smooth losses (like Cross-Entropy) enable fast optimization rates but yield slow square-root consistency bounds, while piecewise-linear losses (like H…

  2. arXiv stat.ML TIER_1 English(EN) · Yutao Zhong ·

    Linear-Core Surrogates: Smooth Loss Functions with Linear Rates for Classification and Structured Prediction

    The choice of loss function in classification involves a fundamental trade-off: smooth losses (like Cross-Entropy) enable fast optimization rates but yield slow square-root consistency bounds, while piecewise-linear losses (like Hinge) offer fast linear consistency rates but suff…