Sobolev Norm Learning Rates for Regularized Least-Squares Algorithms

Authors: Simon Fischer, Ingo Steinwart

JMLR 2020 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical Learning rates for least-squares regression are typically expressed in terms of L2-norms. In this paper we extend these rates to norms stronger than the L2-norm without requiring the regression function to be contained in the hypothesis space. ... Finally, we prove the asymptotic optimality of our results in many cases. Keywords: statistical learning theory, regularized kernel methods, least-squares regression, interpolation norms, uniform convergence, learning rates
Researcher Affiliation Academia Simon Fischer EMAIL Ingo Steinwart EMAIL Institute for Stochastics and Applications Faculty 8: Mathematics and Physics University of Stuttgart 70569 Stuttgart Germany
Pseudocode No The paper describes mathematical proofs and theorems without providing any structured pseudocode or algorithm blocks. It primarily focuses on theoretical derivations.
Open Source Code No The paper does not contain any explicit statements about releasing source code, nor does it provide links to a code repository. The arXiv reference is for a previous version of the paper itself, not for source code.
Open Datasets No The paper discusses theoretical concepts related to learning rates for regularized least-squares algorithms and does not perform experiments on any specific public or open datasets. The mention of 'data set D' is a theoretical concept in the introduction.
Dataset Splits No The paper is theoretical and does not involve empirical experiments with datasets, therefore it does not discuss dataset splits.
Hardware Specification No This paper presents theoretical results and mathematical proofs; it does not describe any experimental setup or specify hardware used for computations.
Software Dependencies No The paper is theoretical and does not describe any implementation details or software dependencies with specific version numbers.
Experiment Setup No The paper focuses on theoretical derivations and proofs of learning rates; it does not include an experimental section with details on hyperparameter values or system-level training settings.