PulseAugur
实时 12:05:41
English(EN) DREG: A Layer-Wise Jacobian Regularization as a General-Purpose Penalty

DREG正则化方法在深度学习中展现出卓越的准确性

研究人员推出了一种名为DREG的逐层雅可比正则化技术,该技术可作为神经网络的通用惩罚。在一项大规模实证研究中,DREG与其他正则化器相比,展现出卓越的准确性,尤其是在数据稀疏和使用Transformer架构中常见的GELU激活函数时。该方法持续优于基线,并在噪声鲁棒性方面排名第二,表明其作为深度学习模型的即插即用解决方案的潜力。 AI

影响 DREG的性能表明,它可以提高深度学习模型的效率和准确性,尤其是在数据稀缺的环境中。

排序理由 该集群包含一篇详细介绍神经网络新正则化技术的论文。

在 arXiv cs.LG 阅读 →

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

DREG正则化方法在深度学习中展现出卓越的准确性

报道来源 [2]

  1. arXiv cs.LG TIER_1 English(EN) · Rowan Martnishn ·

    DREG: A Layer-Wise Jacobian Regularization as a General-Purpose Penalty

    arXiv:2606.23942v1 Announce Type: new Abstract: We present a large-scale empirical study isolating the contributions of the Derivative Regularization penalty (DREG). Across a fully-crossed factorial sweep of 960 experiments spanning 4 activations, 6 regularizers, 8 datasets, and …

  2. arXiv cs.LG TIER_1 English(EN) · Rowan Martnishn ·

    DREG: A Layer-Wise Jacobian Regularization as a General-Purpose Penalty

    We present a large-scale empirical study isolating the contributions of the Derivative Regularization penalty (DREG). Across a fully-crossed factorial sweep of 960 experiments spanning 4 activations, 6 regularizers, 8 datasets, and 5 random seeds, we ask: when, where, and why doe…