Malign Overfitting: Interpolation Can Provably Preclude Invariance

• URL: http://arxiv.org/abs/2211.15724v2
• Date: Wed, 3 Jul 2024 15:40:39 GMT
• Title: Malign Overfitting: Interpolation Can Provably Preclude Invariance
• Authors: Yoav Wald, Gal Yona, Uri Shalit, Yair Carmon,
• Abstract summary: We show that "benign overfitting" in which models generalize well despite interpolating might not favorably extend to settings in which robustness or fairness are desirable. We propose and analyze an algorithm that successfully learns a non-interpolating classifier that is provably invariant.
• Score: 30.776243638012314
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Learned classifiers should often possess certain invariance properties meant to encourage fairness, robustness, or out-of-distribution generalization. However, multiple recent works empirically demonstrate that common invariance-inducing regularizers are ineffective in the over-parameterized regime, in which classifiers perfectly fit (i.e. interpolate) the training data. This suggests that the phenomenon of "benign overfitting", in which models generalize well despite interpolating, might not favorably extend to settings in which robustness or fairness are desirable. In this work we provide a theoretical justification for these observations. We prove that -- even in the simplest of settings -- any interpolating learning rule (with arbitrarily small margin) will not satisfy these invariance properties. We then propose and analyze an algorithm that -- in the same setting -- successfully learns a non-interpolating classifier that is provably invariant. We validate our theoretical observations on simulated data and the Waterbirds dataset.

Related papers

• Equivariance and Invariance Inductive Bias for Learning from Insufficient Data [65.42329520528223]
  We show why insufficient data renders the model more easily biased to the limited training environments that are usually different from testing. We propose a class-wise invariant risk minimization (IRM) that efficiently tackles the challenge of missing environmental annotation in conventional IRM.
  arXiv  Detail & Related papers  (2022-07-25T15:26:19Z)
• On the Strong Correlation Between Model Invariance and Generalization [54.812786542023325]
  Generalization captures a model's ability to classify unseen data. Invariance measures consistency of model predictions on transformations of the data. From a dataset-centric view, we find a certain model's accuracy and invariance linearly correlated on different test sets.
  arXiv  Detail & Related papers  (2022-07-14T17:08:25Z)
• Predicting Out-of-Domain Generalization with Neighborhood Invariance [59.05399533508682]
  We propose a measure of a classifier's output invariance in a local transformation neighborhood. Our measure is simple to calculate, does not depend on the test point's true label, and can be applied even in out-of-domain (OOD) settings. In experiments on benchmarks in image classification, sentiment analysis, and natural language inference, we demonstrate a strong and robust correlation between our measure and actual OOD generalization.
  arXiv  Detail & Related papers  (2022-07-05T14:55:16Z)
• Ensembling over Classifiers: a Bias-Variance Perspective [13.006468721874372]
  We build upon the extension to the bias-variance decomposition by Pfau (2013) in order to gain crucial insights into the behavior of ensembles of classifiers. We show that conditional estimates necessarily incur an irreducible error. Empirically, standard ensembling reducesthe bias, leading us to hypothesize that ensembles of classifiers may perform well in part because of this unexpected reduction.
  arXiv  Detail & Related papers  (2022-06-21T17:46:35Z)
• Equivariance Discovery by Learned Parameter-Sharing [153.41877129746223]
  We study how to discover interpretable equivariances from data. Specifically, we formulate this discovery process as an optimization problem over a model's parameter-sharing schemes. Also, we theoretically analyze the method for Gaussian data and provide a bound on the mean squared gap between the studied discovery scheme and the oracle scheme.
  arXiv  Detail & Related papers  (2022-04-07T17:59:19Z)
• Nonlinear Invariant Risk Minimization: A Causal Approach [5.63479133344366]
  We propose a learning paradigm that enables out-of-distribution generalization in the nonlinear setting. We show identifiability of the data representation up to very simple transformations. Extensive experiments on both synthetic and real-world datasets show that our approach significantly outperforms a variety of baseline methods.
  arXiv  Detail & Related papers  (2021-02-24T15:38:41Z)
• Understanding Double Descent Requires a Fine-Grained Bias-Variance Decomposition [34.235007566913396]
  We describe an interpretable, symmetric decomposition of the variance into terms associated with the labels. We find that the bias decreases monotonically with the network width, but the variance terms exhibit non-monotonic behavior. We also analyze the strikingly rich phenomenology that arises.
  arXiv  Detail & Related papers  (2020-11-04T21:04:02Z)
• Learning Invariances in Neural Networks [51.20867785006147]
  We show how to parameterize a distribution over augmentations and optimize the training loss simultaneously with respect to the network parameters and augmentation parameters. We can recover the correct set and extent of invariances on image classification, regression, segmentation, and molecular property prediction from a large space of augmentations.
  arXiv  Detail & Related papers  (2020-10-22T17:18:48Z)
• What causes the test error? Going beyond bias-variance via ANOVA [21.359033212191218]
  Modern machine learning methods are often overparametrized, allowing adaptation to the data at a fine level. Recent work aimed to understand in greater depth why overparametrization is helpful for generalization. We propose using the analysis of variance (ANOVA) to decompose the variance in the test error in a symmetric way.
  arXiv  Detail & Related papers  (2020-10-11T05:21:13Z)
• Good Classifiers are Abundant in the Interpolating Regime [64.72044662855612]
  We develop a methodology to compute precisely the full distribution of test errors among interpolating classifiers. We find that test errors tend to concentrate around a small typical value $varepsilon*$, which deviates substantially from the test error of worst-case interpolating model. Our results show that the usual style of analysis in statistical learning theory may not be fine-grained enough to capture the good generalization performance observed in practice.
  arXiv  Detail & Related papers  (2020-06-22T21:12:31Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.