Does data interpolation contradict statistical optimality?

We show that learning methods interpolating the training data can achieve optimal rates for the problems of nonparametric regression and prediction with square loss.

stat.ML cs.LG math.ST Mathematics Computer Science

View graph Source

Paper

References (14)

01Overfitting or perfect fitting? Risk bounds for classification and regression rules that interpolateMikhail Belkin, Daniel J. Hsu, P. Mitra2018 · Neural Information Processing Systems · 280 citations In Library

02Rates of Convergence for Nearest Neighbor ClassificationKamalika Chaudhuri, S. Dasgupta2014 · Neural Information Processing Systems · 161 citations In Library

03Empirical Entropy, Minimax Regret and Minimax RiskA. Rakhlin, Karthik Sridharan, A. Tsybakov2013 · arXiv.org · 87 citations In Library

04Introduction to Nonparametric EstimationA. Tsybakov2008 · Springer Series in Statistics · 3.5k citations

05The Elements of Statistical LearningE. Ziegel2003 · Technometrics · 19k citations

06The Hilbert Kernel Regression EstimateL. Devroye, L. Györfi, A. Krzyżak1998 · 30 citations

07Surfaces generated by moving least squares methodsP. Lancaster, K. Salkauskas1981 · 2.7k citations

08A two-dimensional interpolation function for irregularly-spaced dataD. Shepard1968 · ACM National Conference · 5.7k citations

09A Probabilistic Theory of Pattern RecognitionL. Devroye, L. Györfi, G. Lugosi1996 · Stochastic Modelling and Applied Probability · 4.1k citations

10Nonparametric identification and smoothing of data (Local approximation methods)1985 · Nauka, Moscow,

11Nearest neighbor pattern classificationT. Cover, P. Hart1967 · IEEE Transactions on Information Theory · 16k citations

12On Estimating RegressionE. Nadaraya1964 · 4.0k citations

Scroll for more · 2 remaining

Fast Rates for Noisy Interpolation Require Rethinking the Effects of Inductive BiasKonstantin Donhauser, Nicolò Ruggeri, Stefan Stojanovic et al.2022 · International Conference on Machine Learning · 24 citations In Library

The Interpolating Regime: Rethinking Regression in High DimensionsSÉRGIO DE ANDRADE, PAULO2025 · Zenodo (CERN European Organization for Nuclear Research) In Library

Benefit of Interpolation in Nearest Neighbor AlgorithmsYue Xing, Qifan Song, Guang Cheng2022 · arXiv (Cornell University) In Library

Overfitting or perfect fitting? Risk bounds for classification and regression rules that interpolateMikhail Belkin, Daniel Hsu, Partha P. Mitra2018 · arXiv (Cornell University) · 280 citations In Library

Benefit of Interpolation in Nearest Neighbor AlgorithmsYue Xing, Qifan Song, Guang Cheng2022 · SIAM Journal on Mathematics of Data Science In Library

On damage of interpolation to adversarial robustness in regressionJing Peng, Yuhong Yang2026 · arXiv.org · 1 citation In Library

Empirical risk minimization in the interpolating regime with application to neural network learningNicole Mücke, Ingo Steinwart2025 · Machine Learning In Library

Patents near this paper

INTERPOLATING ISOTONIC REGRESSION FOR BINARY CLASSIFICATION OF SPAM, EXPLICIT MATERIAL OR MALWARE USING INTERPOLATION BASED ON A DELAUNAY TRIANGULATIONUS 9,189,752 · GOOGLE INC. · 2015 METHOD AND APPARATUS FOR FAST INTERPOLATION OF MULTI-DIMENSIONAL FUNCTIONS WITH NON-RECTANGULAR DATA SETSUS 7,203,716 · SIMMONDS PRECISION PRODUCTS, INC. · 2007 METHOD OF DISTRIBUTING A RANDOM VARIABLE USING STATISTICALLY CORRECT SPATIAL INTERPOLATION CONTINUOUSLY WITH SPATIALLY INHOMOGENEOUS STATISTICAL CORRELATION VERSUS DISTANCE, STANDARD DEVIATION, AND MEANUS 8,423,328 · INTERNATIONAL BUSINESS MACHINES CORPORATION · 2013 Systems and Methods for Learning Effective Loss Functions EfficientlyUS 11,657,118 · GOOGLE LLC · 2023 SYSTEMS AND METHODS FOR GENERATING INTERPOLATED DATA SETS CONVERGING TO OPTIMIZED RESULTS USING ITERATIVE OVERLAPPING INPUTSUS 8,768,942 · RED HAT, INC. · 2014

Source of record

arXiv

Does data interpolation contradict statistical optimality?

Paper

References (14)

Similar papers

Patents near this paper

Welcome.

Patents near this paper