Optimization for Robustness Evaluation Beyond ℓp Metrics

Hengyue Liang; Buyun Liang; Ying Cui; Tim Mitchell; Ju Sun

doi:10.1109/ICASSP49357.2023.10095871

Optimization for Robustness Evaluation Beyond ℓ_p Metrics

Hengyue Liang, Buyun Liang, Ying Cui, Tim Mitchell, Ju Sun

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

Empirical evaluation of the adversarial robustness of deep learning models involves solving non-trivial constrained optimization problems. Popular numerical algorithms to solve these constrained problems rely predominantly on projected gradient descent (PGD) and mostly handle adversarial perturbations modeled by the ℓ₁, ℓ₂, and ℓ_∞ metrics. In this paper, we introduce a novel algorithmic framework that blends a general-purpose constrained-optimization solver PyGRANSO, With Constraint-Folding (PWCF), to add reliability and generality to robustness evaluation. PWCF 1) finds good-quality solutions without the need of delicate hyperparameter tuning and 2) can handle more general perturbation types, e.g., modeled by general ℓ_p (where p > 0) and perceptual (nonℓ_p) distances, which are inaccessible to existing PGD-based algorithms.

Original language	English (US)
Title of host publication	ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing, Proceedings
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9781728163277
DOIs	https://doi.org/10.1109/ICASSP49357.2023.10095871
State	Published - 2023
Event	48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023 - Rhodes Island, Greece Duration: Jun 4 2023 → Jun 10 2023

Publication series

Name	ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
Volume	2023-June
ISSN (Print)	1520-6149

Conference

Conference	48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023
Country/Territory	Greece
City	Rhodes Island
Period	6/4/23 → 6/10/23

Bibliographical note

Publisher Copyright:
© 2023 IEEE.

Keywords

adversarial attack
adversarial robustness
constrained optimization
deep neural networks
perceptual distance
robustness evaluation

Access

10.1109/ICASSP49357.2023.10095871

OpenUrl availability

Full text

Cite this

Liang, H., Liang, B., Cui, Y., Mitchell, T., & Sun, J. (2023). Optimization for Robustness Evaluation Beyond ℓ_p Metrics. In ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing, Proceedings (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings; Vol. 2023-June). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICASSP49357.2023.10095871

Optimization for Robustness Evaluation Beyond ℓ_p Metrics. / Liang, Hengyue; Liang, Buyun; Cui, Ying et al.
ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing, Proceedings. Institute of Electrical and Electronics Engineers Inc., 2023. (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings; Vol. 2023-June).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Liang, H, Liang, B, Cui, Y, Mitchell, T & Sun, J 2023, Optimization for Robustness Evaluation Beyond ℓ_p Metrics. in ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing, Proceedings. ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, vol. 2023-June, Institute of Electrical and Electronics Engineers Inc., 48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023, Rhodes Island, Greece, 6/4/23. https://doi.org/10.1109/ICASSP49357.2023.10095871

Liang H, Liang B, Cui Y, Mitchell T, Sun J. Optimization for Robustness Evaluation Beyond ℓ_p Metrics. In ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing, Proceedings. Institute of Electrical and Electronics Engineers Inc. 2023. (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings). doi: 10.1109/ICASSP49357.2023.10095871

Liang, Hengyue ; Liang, Buyun ; Cui, Ying et al. / Optimization for Robustness Evaluation Beyond ℓ_p Metrics. ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing, Proceedings. Institute of Electrical and Electronics Engineers Inc., 2023. (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings).

@inproceedings{134dc49935b5452284d3134172f7bfde,

title = "Optimization for Robustness Evaluation Beyond ℓp Metrics",

abstract = "Empirical evaluation of the adversarial robustness of deep learning models involves solving non-trivial constrained optimization problems. Popular numerical algorithms to solve these constrained problems rely predominantly on projected gradient descent (PGD) and mostly handle adversarial perturbations modeled by the ℓ1, ℓ2, and ℓ∞ metrics. In this paper, we introduce a novel algorithmic framework that blends a general-purpose constrained-optimization solver PyGRANSO, With Constraint-Folding (PWCF), to add reliability and generality to robustness evaluation. PWCF 1) finds good-quality solutions without the need of delicate hyperparameter tuning and 2) can handle more general perturbation types, e.g., modeled by general ℓp (where p > 0) and perceptual (nonℓp) distances, which are inaccessible to existing PGD-based algorithms.",

keywords = "adversarial attack, adversarial robustness, constrained optimization, deep neural networks, perceptual distance, robustness evaluation",

author = "Hengyue Liang and Buyun Liang and Ying Cui and Tim Mitchell and Ju Sun",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023 ; Conference date: 04-06-2023 Through 10-06-2023",

year = "2023",

doi = "10.1109/ICASSP49357.2023.10095871",

language = "English (US)",

series = "ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

booktitle = "ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing, Proceedings",

}

TY - GEN

T1 - Optimization for Robustness Evaluation Beyond ℓp Metrics

AU - Liang, Hengyue

AU - Liang, Buyun

AU - Cui, Ying

AU - Mitchell, Tim

AU - Sun, Ju

PY - 2023

Y1 - 2023

N2 - Empirical evaluation of the adversarial robustness of deep learning models involves solving non-trivial constrained optimization problems. Popular numerical algorithms to solve these constrained problems rely predominantly on projected gradient descent (PGD) and mostly handle adversarial perturbations modeled by the ℓ1, ℓ2, and ℓ∞ metrics. In this paper, we introduce a novel algorithmic framework that blends a general-purpose constrained-optimization solver PyGRANSO, With Constraint-Folding (PWCF), to add reliability and generality to robustness evaluation. PWCF 1) finds good-quality solutions without the need of delicate hyperparameter tuning and 2) can handle more general perturbation types, e.g., modeled by general ℓp (where p > 0) and perceptual (nonℓp) distances, which are inaccessible to existing PGD-based algorithms.

AB - Empirical evaluation of the adversarial robustness of deep learning models involves solving non-trivial constrained optimization problems. Popular numerical algorithms to solve these constrained problems rely predominantly on projected gradient descent (PGD) and mostly handle adversarial perturbations modeled by the ℓ1, ℓ2, and ℓ∞ metrics. In this paper, we introduce a novel algorithmic framework that blends a general-purpose constrained-optimization solver PyGRANSO, With Constraint-Folding (PWCF), to add reliability and generality to robustness evaluation. PWCF 1) finds good-quality solutions without the need of delicate hyperparameter tuning and 2) can handle more general perturbation types, e.g., modeled by general ℓp (where p > 0) and perceptual (nonℓp) distances, which are inaccessible to existing PGD-based algorithms.

KW - adversarial attack

KW - adversarial robustness

KW - constrained optimization

KW - deep neural networks

KW - perceptual distance

KW - robustness evaluation

UR - http://www.scopus.com/inward/record.url?scp=85177548914&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85177548914&partnerID=8YFLogxK

U2 - 10.1109/ICASSP49357.2023.10095871

DO - 10.1109/ICASSP49357.2023.10095871

M3 - Conference contribution

AN - SCOPUS:85177548914

T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

BT - ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing, Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023

Y2 - 4 June 2023 through 10 June 2023

ER -