Kernel interpolation generalizes poorly
成果类型：
Article
署名作者：
Li, Yicheng; Zhang, Haobo; Lin, Qian
署名单位：
Tsinghua University
刊物名称：
BIOMETRIKA
ISSN/ISSBN：
0006-3444
DOI：
10.1093/biomet/asad048
发表日期：
2024
页码：
715722
关键词：
摘要：
One of the most interesting problems in the recent renaissance of the studies in kernel regression might be whether kernel interpolation can generalize well, since it may help us understand the 'benign overfitting phenomenon' reported in the literature on deep networks. In this paper, under mild conditions, we show that, for any epsilon>0, the generalization error of kernel interpolation is lower bounded by Omega(n(-epsilon)). In other words, the kernel interpolation generalizes poorly for a large class of kernels. As a direct corollary, we can show that overfitted wide neural networks defined on the sphere generalize poorly.
来源URL：