English
Related papers

Related papers: Simplicity bias and optimization threshold in two-…

200 papers

Deep neural networks are often trained in the over-parametrized regime (i.e. with far more parameters than training examples), and understanding why the training converges to solutions that generalize remains an open problem. Several…

Machine Learning · Statistics 2018-03-23 Hartmut Maennel , Olivier Bousquet , Sylvain Gelly

We study dropout in two-layer neural networks with rectified linear unit (ReLU) activations. Under mild overparametrization and assuming that the limiting kernel can separate the data distribution with a positive margin, we show that…

Machine Learning · Computer Science 2020-10-27 Poorya Mianjy , Raman Arora

This paper studies the problem of training a two-layer ReLU network for binary classification using gradient flow with small initialization. We consider a training dataset with well-separated input vectors: Any pair of input data with the…

Machine Learning · Computer Science 2024-03-26 Hancheng Min , Enrique Mallada , René Vidal

There has been a long history of works showing that neural networks have hard time extrapolating beyond the training set. A recent study by Balestriero et al. (2021) challenges this view: defining interpolation as the state of belonging to…

Machine Learning · Computer Science 2022-07-19 Laurent Bonnasse-Gahot

In deep learning, often the training process finds an interpolator (a solution with 0 training loss), but the test loss is still low. This phenomenon, known as benign overfitting, is a major mystery that received a lot of recent attention.…

Machine Learning · Computer Science 2023-05-29 Mo Zhou , Rong Ge

With the motive of training all the parameters of a neural network, we study why and when one can achieve this by iteratively creating, training, and combining randomly selected subnetworks. Such scenarios have either implicitly or…

Machine Learning · Computer Science 2022-08-15 Fangshuo Liao , Anastasios Kyrillidis

The optimization of neural networks under weight decay remains poorly understood from a theoretical standpoint. While weight decay is standard practice in modern training procedures, most theoretical analyses focus on unregularized…

Machine Learning · Computer Science 2025-05-29 Etienne Boursier , Matthew Bowditch , Matthias Englert , Ranko Lazic

The ability of overparameterized deep networks to interpolate noisy data, while at the same time showing good generalization performance, has been recently characterized in terms of the double descent curve for the test error. Common…

Machine Learning · Computer Science 2023-04-11 Matteo Gamba , Erik Englesson , Mårten Björkman , Hossein Azizpour

Neural network training is usually accomplished by solving a non-convex optimization problem using stochastic gradient descent. Although one optimizes over the networks parameters, the main loss function generally only depends on the…

Machine Learning · Computer Science 2023-02-10 Julius Berner , Dennis Elbrächter , Philipp Grohs

We study the generalization error of functions that interpolate prescribed data points and are selected by minimizing a weighted norm. Under natural and general conditions, we prove that both the interpolants and their generalization errors…

Numerical Analysis · Mathematics 2021-02-11 Weilin Li

The recent success of neural network models has shone light on a rather surprising statistical phenomenon: statistical models that perfectly fit noisy data can generalize well to unseen test data. Understanding this phenomenon of…

Machine Learning · Statistics 2022-09-13 Niladri S. Chatterji , Philip M. Long , Peter L. Bartlett

Neural networks often exhibit simplicity bias, favoring simpler features over more complex ones, even when both are equally predictive. We introduce a novel method called imbalanced label coupling to explore and extend this simplicity bias…

Machine Learning · Computer Science 2024-10-23 Zhehang Du

Neural networks trained with SGD were recently shown to rely preferentially on linearly-predictive features and can ignore complex, equally-predictive ones. This simplicity bias can explain their lack of robustness out of distribution…

Machine Learning · Computer Science 2022-09-13 Damien Teney , Ehsan Abbasnejad , Simon Lucey , Anton van den Hengel

A continuing mystery in understanding the empirical success of deep neural networks is their ability to achieve zero training error and generalize well, even when the training data is noisy and there are more parameters than data points. We…

Machine Learning · Computer Science 2019-09-10 Vidya Muthukumar , Kailas Vodrahalli , Vignesh Subramanian , Anant Sahai

Neural networks can be trained to solve regression problems by using gradient-based methods to minimize the square loss. However, practitioners often prefer to reformulate regression as a classification problem, observing that training on…

Machine Learning · Computer Science 2023-03-02 Lawrence Stewart , Francis Bach , Quentin Berthet , Jean-Philippe Vert

Algorithmic stability is a classical framework for analyzing the generalization error of learning algorithms. It predicts that an algorithm has small generalization error if it is insensitive to small perturbations in the training set such…

Machine Learning · Computer Science 2026-02-17 Ouns El Harzli , Yoonsoo Nam , Ilja Kuzborskij , Bernardo Cuenca Grau , Ard A. Louis

We consider the problem of generalization of arbitrarily overparameterized two-layer ReLU Neural Networks with univariate input. Recent work showed that under square loss, flat solutions (motivated by flat / stable minima and Edge of…

Machine Learning · Computer Science 2025-12-02 Dan Qiao , Yu-Xiang Wang

It has been observed in practical applications and in theoretical analysis that over-parametrization helps to find good minima in neural network training. Similarly, in this article we study widening and deepening neural networks by a…

Numerical Analysis · Mathematics 2020-02-06 G. Welper

When optimizing over-parameterized models, such as deep neural networks, a large set of parameters can achieve zero training error. In such cases, the choice of the optimization algorithm and its respective hyper-parameters introduces…

Machine Learning · Computer Science 2019-12-06 Gauthier Gidel , Francis Bach , Simon Lacoste-Julien

We bound the excess risk of interpolating deep linear networks trained using gradient flow. In a setting previously used to establish risk bounds for the minimum $\ell_2$-norm interpolant, we show that randomly initialized deep linear…

Machine Learning · Computer Science 2023-02-08 Niladri S. Chatterji , Philip M. Long