A minimax optimal control approach for robust neural ODEs

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Cipriani, Cristina, Scagliotti, Alessandro, Wöhrer, Tobias
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913291505238016
author Cipriani, Cristina
Scagliotti, Alessandro
Wöhrer, Tobias
author_facet Cipriani, Cristina
Scagliotti, Alessandro
Wöhrer, Tobias
contents In this paper, we address the adversarial training of neural ODEs from a robust control perspective. This is an alternative to the classical training via empirical risk minimization, and it is widely used to enforce reliable outcomes for input perturbations. Neural ODEs allow the interpretation of deep neural networks as discretizations of control systems, unlocking powerful tools from control theory for the development and the understanding of machine learning. In this specific case, we formulate the adversarial training with perturbed data as a minimax optimal control problem, for which we derive first order optimality conditions in the form of Pontryagin's Maximum Principle. We provide a novel interpretation of robust training leading to an alternative weighted technique, which we test on a low-dimensional classification task.
format Preprint
id arxiv_https___arxiv_org_abs_2310_17584
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle A minimax optimal control approach for robust neural ODEs
Cipriani, Cristina
Scagliotti, Alessandro
Wöhrer, Tobias
Optimization and Control
Machine Learning
Systems and Control
In this paper, we address the adversarial training of neural ODEs from a robust control perspective. This is an alternative to the classical training via empirical risk minimization, and it is widely used to enforce reliable outcomes for input perturbations. Neural ODEs allow the interpretation of deep neural networks as discretizations of control systems, unlocking powerful tools from control theory for the development and the understanding of machine learning. In this specific case, we formulate the adversarial training with perturbed data as a minimax optimal control problem, for which we derive first order optimality conditions in the form of Pontryagin's Maximum Principle. We provide a novel interpretation of robust training leading to an alternative weighted technique, which we test on a low-dimensional classification task.
title A minimax optimal control approach for robust neural ODEs
topic Optimization and Control
Machine Learning
Systems and Control
url https://arxiv.org/abs/2310.17584