Saved in:
Bibliographic Details
Main Authors: Tsigler, Alexander, Chamon, Luiz F. O., Frei, Spencer, Bartlett, Peter L.
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2503.07966
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909533637443584
author Tsigler, Alexander
Chamon, Luiz F. O.
Frei, Spencer
Bartlett, Peter L.
author_facet Tsigler, Alexander
Chamon, Luiz F. O.
Frei, Spencer
Bartlett, Peter L.
contents In this work, we investigate the behavior of ridge regression in an overparameterized binary classification task. We assume examples are drawn from (anisotropic) class-conditional cluster distributions with opposing means and we allow for the training labels to have a constant level of label-flipping noise. We characterize the classification error achieved by ridge regression under the assumption that the covariance matrix of the cluster distribution has a high effective rank in the tail. We show that ridge regression has qualitatively different behavior depending on the scale of the cluster mean vector and its interaction with the covariance matrix of the cluster distributions. In regimes where the scale is very large, the conditions that allow for benign overfitting turn out to be the same as those for the regression task. We additionally provide insights into how the introduction of label noise affects the behavior of the minimum norm interpolator (MNI). The optimal classifier in this setting is a linear transformation of the cluster mean vector and in the noiseless setting the MNI approximately learns this transformation. On the other hand, the introduction of label noise can significantly change the geometry of the solution while preserving the same qualitative behavior.
format Preprint
id arxiv_https___arxiv_org_abs_2503_07966
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Benign Overfitting and the Geometry of the Ridge Regression Solution in Binary Classification
Tsigler, Alexander
Chamon, Luiz F. O.
Frei, Spencer
Bartlett, Peter L.
Machine Learning
In this work, we investigate the behavior of ridge regression in an overparameterized binary classification task. We assume examples are drawn from (anisotropic) class-conditional cluster distributions with opposing means and we allow for the training labels to have a constant level of label-flipping noise. We characterize the classification error achieved by ridge regression under the assumption that the covariance matrix of the cluster distribution has a high effective rank in the tail. We show that ridge regression has qualitatively different behavior depending on the scale of the cluster mean vector and its interaction with the covariance matrix of the cluster distributions. In regimes where the scale is very large, the conditions that allow for benign overfitting turn out to be the same as those for the regression task. We additionally provide insights into how the introduction of label noise affects the behavior of the minimum norm interpolator (MNI). The optimal classifier in this setting is a linear transformation of the cluster mean vector and in the noiseless setting the MNI approximately learns this transformation. On the other hand, the introduction of label noise can significantly change the geometry of the solution while preserving the same qualitative behavior.
title Benign Overfitting and the Geometry of the Ridge Regression Solution in Binary Classification
topic Machine Learning
url https://arxiv.org/abs/2503.07966