Matches in SemOpenAlex for { <https://semopenalex.org/work/W2899853676> ?p ?o ?g. }
Showing items 1 to 80 of
80
with 100 items per page.
- W2899853676 abstract "We study the problem of finding a universal (image-agnostic) perturbation to fool machine learning (ML) classifiers (e.g., neural nets, decision tress) in the hard-label black-box setting. Recent work in adversarial ML in the white-box setting (model parameters are known) has shown that many state-of-the-art image classifiers are vulnerable to universal adversarial perturbations: a fixed human-imperceptible perturbation that, when added to any image, causes it to be misclassified with high probability Kurakin et al. [2016], Szegedy et al. [2013], Chen et al. [2017a], Carlini and Wagner [2017]. This paper considers a more practical and challenging problem of finding such universal perturbations in an obscure (or black-box) setting. More specifically, we use zeroth order optimization algorithms to find such a universal adversarial perturbation when no model information is revealed-except that the attacker can make queries to probe the classifier. We further relax the assumption that the output of a query is continuous valued confidence scores for all the classes and consider the case where the output is a hard-label decision. Surprisingly, we found that even in these extremely obscure regimes, state-of-the-art ML classifiers can be fooled with a very high probability just by adding a single human-imperceptible image perturbation to any natural image. The surprising existence of universal perturbations in a hard-label black-box setting raises serious security concerns with the existence of a universal noise vector that adversaries can possibly exploit to break a classifier on most natural images." @default.
- W2899853676 created "2018-11-16" @default.
- W2899853676 creator A5041470575 @default.
- W2899853676 creator A5050402276 @default.
- W2899853676 date "2018-11-09" @default.
- W2899853676 modified "2023-09-27" @default.
- W2899853676 title "Universal Decision-Based Black-Box Perturbations: Breaking Security-Through-Obscurity Defenses." @default.
- W2899853676 cites W2460937040 @default.
- W2899853676 cites W2781152419 @default.
- W2899853676 cites W2890569370 @default.
- W2899853676 cites W2903939901 @default.
- W2899853676 cites W2908776022 @default.
- W2899853676 cites W2963070423 @default.
- W2899853676 cites W2963243330 @default.
- W2899853676 cites W2963612069 @default.
- W2899853676 cites W2963855547 @default.
- W2899853676 cites W2963857521 @default.
- W2899853676 cites W2964153729 @default.
- W2899853676 cites W2964171870 @default.
- W2899853676 cites W2964205597 @default.
- W2899853676 hasPublicationYear "2018" @default.
- W2899853676 type Work @default.
- W2899853676 sameAs 2899853676 @default.
- W2899853676 citedByCount "2" @default.
- W2899853676 countsByYear W28998536762018 @default.
- W2899853676 countsByYear W28998536762019 @default.
- W2899853676 crossrefType "posted-content" @default.
- W2899853676 hasAuthorship W2899853676A5041470575 @default.
- W2899853676 hasAuthorship W2899853676A5050402276 @default.
- W2899853676 hasConcept C11413529 @default.
- W2899853676 hasConcept C119857082 @default.
- W2899853676 hasConcept C121332964 @default.
- W2899853676 hasConcept C154945302 @default.
- W2899853676 hasConcept C165696696 @default.
- W2899853676 hasConcept C177918212 @default.
- W2899853676 hasConcept C37736160 @default.
- W2899853676 hasConcept C38652104 @default.
- W2899853676 hasConcept C41008148 @default.
- W2899853676 hasConcept C62520636 @default.
- W2899853676 hasConcept C94966114 @default.
- W2899853676 hasConcept C95623464 @default.
- W2899853676 hasConceptScore W2899853676C11413529 @default.
- W2899853676 hasConceptScore W2899853676C119857082 @default.
- W2899853676 hasConceptScore W2899853676C121332964 @default.
- W2899853676 hasConceptScore W2899853676C154945302 @default.
- W2899853676 hasConceptScore W2899853676C165696696 @default.
- W2899853676 hasConceptScore W2899853676C177918212 @default.
- W2899853676 hasConceptScore W2899853676C37736160 @default.
- W2899853676 hasConceptScore W2899853676C38652104 @default.
- W2899853676 hasConceptScore W2899853676C41008148 @default.
- W2899853676 hasConceptScore W2899853676C62520636 @default.
- W2899853676 hasConceptScore W2899853676C94966114 @default.
- W2899853676 hasConceptScore W2899853676C95623464 @default.
- W2899853676 hasLocation W28998536761 @default.
- W2899853676 hasOpenAccess W2899853676 @default.
- W2899853676 hasPrimaryLocation W28998536761 @default.
- W2899853676 hasRelatedWork W1551944411 @default.
- W2899853676 hasRelatedWork W1891967673 @default.
- W2899853676 hasRelatedWork W2031633693 @default.
- W2899853676 hasRelatedWork W2406297787 @default.
- W2899853676 hasRelatedWork W2514337832 @default.
- W2899853676 hasRelatedWork W2807539765 @default.
- W2899853676 hasRelatedWork W2889529480 @default.
- W2899853676 hasRelatedWork W2892179671 @default.
- W2899853676 hasRelatedWork W2901276153 @default.
- W2899853676 hasRelatedWork W2951538235 @default.
- W2899853676 hasRelatedWork W2952049331 @default.
- W2899853676 hasRelatedWork W2969992695 @default.
- W2899853676 hasRelatedWork W3007802865 @default.
- W2899853676 hasRelatedWork W3025116310 @default.
- W2899853676 hasRelatedWork W3091701562 @default.
- W2899853676 hasRelatedWork W3126570451 @default.
- W2899853676 hasRelatedWork W3127395865 @default.
- W2899853676 hasRelatedWork W3131646897 @default.
- W2899853676 hasRelatedWork W3166408207 @default.
- W2899853676 hasRelatedWork W75861488 @default.
- W2899853676 isParatext "false" @default.
- W2899853676 isRetracted "false" @default.
- W2899853676 magId "2899853676" @default.
- W2899853676 workType "article" @default.