Analysis of classifiers' robustness to adversarial perturbations

DOI10.1007/s10994-017-5663-3zbMath1462.62383arXiv1502.02590OpenAlexW2963739340WikidataQ62556705 ScholiaQ62556705MaRDI QIDQ1640392

Pascal Frossard, Alhussein Fawzi, Omar Fawzi

Publication date: 14 June 2018

Published in: Machine Learning (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/1502.02590

zbMATH Keywords

instabilityrandom noiseadversarial examplesclassification robustnessdeep networks

Mathematics Subject Classification ID

Nonparametric robustness (62G35) Classification and discrimination; cluster analysis (statistical aspects) (62H30) Neural nets and related approaches to inference from stochastic processes (62M45)

Related Items

Black-box adversarial attacks by manipulating image attributes ⋮ Achieving adversarial robustness via sparsity ⋮ Achieving Adversarial Robustness Requires An Active Teacher ⋮ On the regularized risk of distributionally robust learning over deep neural networks ⋮ A robust outlier control framework for classification designed with family of homotopy loss function ⋮ Metrics and methods for robustness evaluation of neural networks with generative models ⋮ Adversarial Robustness of Sparse Local Lipschitz Predictors ⋮ Adversarial classification via distributional robustness with Wasserstein ambiguity ⋮ Spanning attack: reinforce black-box attacks with unlabeled data ⋮ Deep Convolutional Neural Networks for Image Classification: A Comprehensive Review ⋮ Adversarial noise attacks of deep learning architectures: stability analysis via sparse-modeled signals ⋮ Unnamed Item ⋮ Unnamed Item ⋮ Compositional falsification of cyber-physical systems with machine learning components ⋮ Re-thinking model robustness from stability: a new insight to defend adversarial examples

Uses Software

• LIBSVM
• LIBLINEAR
• SIFT
• CIFAR
• DeepFool

Cites Work

• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Error bounds of constrained quadratic functions and piecewise affine inequality systems
• New fractional error bounds for polynomial systems with applications to Hölderian stability in optimization and spectral theory of tensors
• Error bounds for analytic systems and their applications
• Error bounds in mathematical programming
• Learning to classify with missing and corrupted features
• Extension of Hoffman’s Error Bound to Polynomial Systems
• On the posterior-probability estimate of the error rate of nonparametric classification rules
• 10.1162/153244302760200704
• 10.1162/153244303321897726