Uniform consistency of cross-validation estimators for high-dimensional ridge regression

P Patil, Y Wei, A Rinaldo… - … Conference on Artificial …, 2021 - proceedings.mlr.press
We examine generalized and leave-one-out cross-validation for ridge regression in a
proportional asymptotic framework where the dimension of the feature space grows …

On the interplay between noise and curvature and its effect on optimization and generalization

V Thomas, F Pedregosa… - International …, 2020 - proceedings.mlr.press
The speed at which one can minimize an expected loss using stochastic methods depends
on two properties: the curvature of the loss and the variance of the gradients. While most …

Failures and Successes of Cross-Validation for Early-Stopped Gradient Descent

P Patil, Y Wu, R Tibshirani - International Conference on …, 2024 - proceedings.mlr.press
We analyze the statistical properties of generalized cross-validation (GCV) and leave-one-
out cross-validation (LOOCV) applied to early-stopped gradient descent (GD) in high …

Hypothesis transfer learning with surrogate classification losses: Generalization bounds through algorithmic stability

A Aghbalou, G Staerman - International Conference on …, 2023 - proceedings.mlr.press
Hypothesis transfer learning (HTL) contrasts domain adaptation by allowing for a previous
task leverage, named the source, into a new one, the target, without requiring access to the …

Leave-one-out cross-validation for Bayesian model comparison in large data

M Magnusson, A Vehtari, J Jonasson… - International …, 2020 - proceedings.mlr.press
Recently, new methods for model assessment, based on subsampling and posterior
approximations, have been proposed for scaling leave-one-out cross-validation (LOO-CV) to …

Asymptotics of cross-validation

M Austern, W Zhou - arxiv preprint arxiv:2001.11111, 2020 - arxiv.org
Cross validation is a central tool in evaluating the performance of machine learning and
statistical models. However, despite its ubiquitous role, its theoretical properties are still not …

Subsample ridge ensembles: Equivalences and generalized cross-validation

JH Du, P Patil, AK Kuchibhotla - arxiv preprint arxiv:2304.13016, 2023 - arxiv.org
We study subsampling-based ridge ensembles in the proportional asymptotics regime,
where the feature size grows proportionally with the sample size such that their ratio …

Iterative approximate cross-validation

Y Luo, Z Ren, R Barber - International Conference on …, 2023 - proceedings.mlr.press
Cross-validation (CV) is one of the most popular tools for assessing and selecting predictive
models. However, standard CV suffers from high computational cost when the number of …

Approximate cross-validation: Guarantees for model assessment and selection

A Wilson, M Kasy, L Mackey - International conference on …, 2020 - proceedings.mlr.press
Cross-validation (CV) is a popular approach for assessing and selecting predictive models.
However, when the number of folds is large, CV suffers from a need to repeatedly refit a …

Is Cross-Validation the Gold Standard to Evaluate Model Performance?

G Iyengar, H Lam, T Wang - arxiv preprint arxiv:2407.02754, 2024 - arxiv.org
Cross-Validation (CV) is the default choice for evaluating the performance of machine
learning models. Despite its wide usage, their statistical benefits have remained half …