Salvato in:
Dettagli Bibliografici
Autori principali: Maurya, Deepak, Barik, Adarsh, Honorio, Jean
Natura: Preprint
Pubblicazione: 2022
Soggetti:
Accesso online:https://arxiv.org/abs/2208.09449
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910709364817920
author Maurya, Deepak
Barik, Adarsh
Honorio, Jean
author_facet Maurya, Deepak
Barik, Adarsh
Honorio, Jean
contents In this work, we propose a robust framework that employs adversarially robust training to safeguard the ML models against perturbed testing data. Our contributions can be seen from both computational and statistical perspectives. Firstly, from a computational/optimization point of view, we derive the ready-to-use exact solution for several widely used loss functions with a variety of norm constraints on adversarial perturbation for various supervised and unsupervised ML problems, including regression, classification, two-layer neural networks, graphical models, and matrix completion. The solutions are either in closed-form, or an easily tractable optimization problem such as 1-D convex optimization, semidefinite programming, difference of convex programming or a sorting-based algorithm. Secondly, from statistical/generalization viewpoint, using some of these results, we derive novel bounds of the adversarial Rademacher complexity for various problems, which entails new generalization bounds. Thirdly, we perform some sanity-check experiments on real-world datasets for supervised problems such as regression and classification, as well as for unsupervised problems such as matrix completion and learning graphical models, with very little computational overhead.
format Preprint
id arxiv_https___arxiv_org_abs_2208_09449
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle A Novel Plug-and-Play Approach for Adversarially Robust Generalization
Maurya, Deepak
Barik, Adarsh
Honorio, Jean
Machine Learning
In this work, we propose a robust framework that employs adversarially robust training to safeguard the ML models against perturbed testing data. Our contributions can be seen from both computational and statistical perspectives. Firstly, from a computational/optimization point of view, we derive the ready-to-use exact solution for several widely used loss functions with a variety of norm constraints on adversarial perturbation for various supervised and unsupervised ML problems, including regression, classification, two-layer neural networks, graphical models, and matrix completion. The solutions are either in closed-form, or an easily tractable optimization problem such as 1-D convex optimization, semidefinite programming, difference of convex programming or a sorting-based algorithm. Secondly, from statistical/generalization viewpoint, using some of these results, we derive novel bounds of the adversarial Rademacher complexity for various problems, which entails new generalization bounds. Thirdly, we perform some sanity-check experiments on real-world datasets for supervised problems such as regression and classification, as well as for unsupervised problems such as matrix completion and learning graphical models, with very little computational overhead.
title A Novel Plug-and-Play Approach for Adversarially Robust Generalization
topic Machine Learning
url https://arxiv.org/abs/2208.09449