Improving Fast Minimum-Norm Attacks with Hyperparameter Optimization
Giuseppe FlorisFirst
;Raffaele Mura;Luca Scionis;Giorgio Piras
;Maura Pintor;Ambra DemontisPenultimate
;Battista BiggioLast
2023-01-01
Abstract
Evaluating the adversarial robustness of machine-learning models using gradient-based attacks is challenging. In this work, we show that hyperparameter optimization can improve fast minimum-norm attacks by automating the selection of the loss function, the optimizer, and the step-size scheduler, along with the corresponding hyperparameters. Our extensive evaluation involving several robust models demonstrates the improved efficacy of fast minimum-norm attacks when hyped up with hyperparameter optimization. We release our open-source code at https://github.com/pralab/HO-FMN.File | Size | Format | |
---|---|---|---|
ES2023-164 (1).pdf Solo gestori archivio
Type: versione editoriale
Size 1.69 MB
Format Adobe PDF
|
1.69 MB | Adobe PDF | & nbsp; View / Open Request a copy |
2310.08177.pdf open access
Type: versione pre-print
Size 443.53 kB
Format Adobe PDF
|
443.53 kB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.