跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2604.23212 · v2

Learning Curves and Prediction Persistence to Under-Regularization of Spectral Algorithms in Large Dimensions

首次提交 2026-04-25 16:38
版本更新 2026-11-19 09:29
雷达观测 2026-11-20 10:50
arXiv ↗PDF ↗

摘要

We study how prediction risk changes as regularization is reduced for spectral algorithms when the sample size nn and dimension dd satisfy n≍dγn\asymp d^γ for fixed γ>0γ>0. For inner-product kernels on the sphere, under a source condition with smoothness s≥0s\geq0 and a signal nondegeneracy condition, we characterize the conditional excess risk up to constant factors at each fixed admissible scale λ≍d−uλ\asymp d^{-u}, u>0u>0, and at interpolation. The resulting risk orders describe the effects of source smoothness, filter qualification, and transitions between spectral blocks. They identify when the conditional excess risk retains the minimax risk order as regularization is weakened from an optimal scale to interpolation, and when an interval before interpolation exhibits a polynomially larger risk. We also compare the kernel risk with a population sequence risk for the same spectral filter, using a deterministic regularization scale bounded below by order n−1n^{-1}. Finally, for noninteger γγ, we extend the risk characterization for kernel ridge regression to kernels on general domains under spectral scaling, hyper-contractivity, and tail diagonal concentration conditions, together with the source and signal conditions.

AI 论文解读

基于摘要 · 非全文审阅

此版本尚未生成 AI 解读。解读由后台任务生成并按论文版本保存,登录后共享已有结果。

版本差异

本站首次收录为 v2,暂无更早版本可对照。

观测到的版本

相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。

v2 · 2026-11-19 09:29 · 最新观测版本

Learning Curves and Prediction Persistence to Under-Regularization of Spectral Algorithms in Large Dimensions

We study how prediction risk changes as regularization is reduced for spectral algorithms when the sample size nn and dimension dd satisfy n≍dγn\asymp d^γ for fixed γ>0γ>0. For inner-product kernels on the sphere, under a source condition with smoothness s≥0s\geq0 and a signal nondegeneracy condition, we characterize the conditional excess risk up to constant factors at each fixed admissible scale λ≍d−uλ\asymp d^{-u}, u>0u>0, and at interpolation. The resulting risk orders describe the effects of source smoothness, filter qualification, and transitions between spectral blocks. They identify when the conditional excess risk retains the minimax risk order as regularization is weakened from an optimal scale to interpolation, and when an interval before interpolation exhibits a polynomially larger risk. We also compare the kernel risk with a population sequence risk for the same spectral filter, using a deterministic regularization scale bounded below by order n−1n^{-1}. Finally, for noninteger γγ, we extend the risk characterization for kernel ridge regression to kernels on general domains under spectral scaling, hyper-contractivity, and tail diagonal concentration conditions, together with the source and signal conditions.

查看此版本解读 →查看此版本原文 ↗