Generalization Error Analysis of Neural Networks with Gradient Based Regularization

Generalization Error Analysis of Neural Networks with Gradient Based Regularization

Year:    2022

Author:    Lingfeng Li, Xue-Cheng Tai, Jiang Yang

Communications in Computational Physics, Vol. 32 (2022), Iss. 4 : pp. 1007–1038

Abstract

In this work, we study gradient-based regularization methods for neural networks. We mainly focus on two regularization methods: the total variation and the Tikhonov regularization. Adding the regularization term to the training loss is equivalent to using neural networks to solve some variational problems, mostly in high dimensions in practical applications. We introduce a general framework to analyze the error between neural network solutions and true solutions to variational problems. The error consists of three parts: the approximation errors of neural networks, the quadrature errors of numerical integration, and the optimization error. We also apply the proposed framework to two-layer networks to derive a priori error estimate when the true solution belongs to the so-called Barron space. Moreover, we conduct some numerical experiments to show that neural networks can solve corresponding variational problems sufficiently well. The networks with gradient-based regularization are much more robust in image applications.

You do not have full access to this article.

Already a Subscriber? Sign in as an individual or via your institution

Journal Article Details

Publisher Name:    Global Science Press

Language:    English

DOI:    https://doi.org/10.4208/cicp.OA-2021-0211

Communications in Computational Physics, Vol. 32 (2022), Iss. 4 : pp. 1007–1038

Published online:    2022-01

AMS Subject Headings:    Global Science Press

Copyright:    COPYRIGHT: © Global Science Press

Pages:    32

Keywords:    Machine learning regularization generalization error image classification.

Author Details

Lingfeng Li

Xue-Cheng Tai

Jiang Yang

  1. A new method to compute the blood flow equations using the physics-informed neural operator

    Li, Lingfeng | Tai, Xue-Cheng | Chan, Raymond Hon-Fu

    Journal of Computational Physics, Vol. 519 (2024), Iss. P.113380

    https://doi.org/10.1016/j.jcp.2024.113380 [Citations: 0]
  2. Federated Morozov Regularization for Shortcut Learning in Privacy Preserving Learning with Watermarked Image Data

    Ling, Tao | Shi, Siping | Wang, Hao | Hu, Chuang | Wang, Dan

    Proceedings of the 32nd ACM International Conference on Multimedia, (2024), P.4899

    https://doi.org/10.1145/3664647.3681480 [Citations: 0]