arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2408.15784cs.LGmath.STstat.MLstat.TH
收藏

加权神经表示的隐式正则化路径

Implicit Regularization Paths of Weighted Neural Representations

Carnegie Mellon University(卡内基梅隆大学);University of California, Berkeley(加利福尼亚大学伯克利分校)

Jin-Hong Du, Pratik Patil

AI总结:

研究了预训练特征加权引发的隐式正则化效应,推导出连接不同加权矩阵与岭正则化水平的等价路径,证实了相关猜想,并据此开发出应用于多个模型和数据集的高效交叉验证调优方法。

摘要:

我们研究了预训练特征的(观测)加权所引发的隐式正则化效应。对于算子范数有界且相对于(归一化)迹泛函无穷小自由的权重矩阵和特征矩阵,我们推导出了连接不同加权矩阵与岭正则化水平的等价路径。具体而言,我们证明了在同一路径上基于加权特征训练的岭估计量,在针对范数有界的测试向量进行评估时,具有渐近等价性。这些路径可以解释为匹配了由加权特征拟合的岭估计量的有效自由度。对于无放回子采样的特殊情形,我们的结果适用于独立采样的随机特征和核特征,并证实了作者在 Patil et al. 中关于此类路径存在的近期猜想(猜想7和8)。我们还提出了加权估计量集成(ensembles)的加性风险分解,并证明了当集成规模趋于无穷大时,风险沿这些路径是等价的。作为路径等价性的一个实际结果,我们开发了一种用于调优的高效交叉验证方法,并将其应用于跨多个模型(例如,ResNet-50)和数据集(例如,CIFAR-100)的子采样预训练表示。

英文摘要:

We study the implicit regularization effects induced by (observation) weighting of pretrained features. For weight and feature matrices of bounded operator norms that are infinitesimally free with respect to (normalized) trace functionals, we derive equivalence paths connecting different weighting matrices and ridge regularization levels. Specifically, we show that ridge estimators trained on weighted features along the same path are asymptotically equivalent when evaluated against test vectors of bounded norms. These paths can be interpreted as matching the effective degrees of freedom of ridge estimators fitted with weighted features. For the special case of subsampling without replacement, our results apply to independently sampled random features and kernel features and confirm recent conjectures (Conjectures 7 and 8) of the authors on the existence of such paths in Patil et al. We also present an additive risk decomposition for ensembles of weighted estimators and show that the risks are equivalent along the paths when the ensemble size goes to infinity. As a practical consequence of the path equivalences, we develop an efficient cross-validation method for tuning and apply it to subsampled pretrained representations across several models (e.g., ResNet-50) and datasets (e.g., CIFAR-100).

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑