On the Optimality of Misspecified Kernel Ridge Regression

Authors: Haobo Zhang, Yicheng Li, Weihao Lu, Qian Lin

ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In our experiments, we aim to verify that for functions f ρ [H]s but not in L , the KRR estimator can still achieve the convergence rate n sβ sβ+1 . We show the results for both Sobolev RKHS, and general RKHS with uniformly bounded eigenfunctions mentioned in Remark 3.3. We numerically compute the generalization error ˆf f L2 by Simpson s formula with N n testing points. For each n, we repeat the experiments 50 times and present the average generalization error as well as the region within one standard deviation.
Researcher Affiliation Collaboration 1Center for Statistical Science, Department of Industrial Engineering, Tsinghua University 2Beijing Academy of Artificial Intelligence, Beijing, China.
Pseudocode No The paper provides a 'Sketch of proof' which describes the mathematical steps but does not include any structured pseudocode or algorithm blocks.
Open Source Code No The paper does not contain any statement about making its source code publicly available or providing a link to a code repository for its methodology.
Open Datasets No The paper generates synthetic data based on mathematical functions and distributions (e.g., 'x U[0, 1]') but does not refer to or provide access information for a publicly available or open dataset.
Dataset Splits Yes We also did another experiment that used cross validation to choose the regularization parameter. Figure 3 in Appendix F shows a similar result as Figure 1. In Figure 3 (b), we use 5-fold cross validation to choose the regularization parameter in KRR and present the logarithmic errors and sample sizes.
Hardware Specification No The paper does not specify any details regarding the hardware (e.g., CPU, GPU models, memory) used for conducting the experiments.
Software Dependencies No The paper does not list specific versions for any software dependencies, libraries, or frameworks used in the experiments.
Experiment Setup Yes We choose the regularization parameter as λ = cn β sβ+1 = cn 10/9 for a fixed c. We try different values of c, Figure 1 presents the convergence curve under the best choice of c.