We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Kernel interpolation generalizes poorly.
- Authors
Li, Yicheng; Zhang, Haobo; Lin, Qian
- Abstract
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 ε > 0 , the generalization error of kernel interpolation is lower bounded by Ω (n − ε) . 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.
- Subjects
INTERPOLATION; GENERALIZATION; RENAISSANCE
- Publication
Biometrika, 2024, Vol 111, Issue 2, p715
- ISSN
0006-3444
- Publication type
Article
- DOI
10.1093/biomet/asad048