FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3128235134> ?p ?o ?g. }

• W3128235134 abstract "Adversarial attacks play an essential role in understanding deep neural network predictions and improving their robustness. Existing attack methods aim to deceive convolutional neural network (CNN)-based classifiers by manipulating RGB images that are fed directly to the classifiers. However, these approaches typically neglect the influence of the camera optics and image processing pipeline (ISP) that produce the network inputs. ISPs transform RAW measurements to RGB images and traditionally are assumed to preserve adversarial patterns. However, these low-level pipelines can, in fact, destroy, introduce or amplify adversarial patterns that can deceive a downstream detector. As a result, optimized patterns can become adversarial for the classifier after being transformed by a certain camera ISP and optic but not for others. In this work, we examine and develop such an attack that deceives a specific camera ISP while leaving others intact, using the same down-stream classifier. We frame camera-specific attacks as a multi-task optimization problem, relying on a differentiable approximation for the ISP itself. We validate the proposed method using recent state-of-the-art automotive hardware ISPs, achieving 92% fooling rate when attacking a specific ISP. We demonstrate physical optics attacks with 90% fooling rate for a specific camera lenses." @default.
• W3128235134 created "2021-02-15" @default.
• W3128235134 creator A5059313827 @default.
• W3128235134 creator A5060060011 @default.
• W3128235134 creator A5085652556 @default.
• W3128235134 date "2021-02-07" @default.
• W3128235134 modified "2023-09-25" @default.
• W3128235134 title "Adversarial Imaging Pipelines" @default.
• W3128235134 cites W1878947057 @default.
• W3128235134 cites W1901129140 @default.
• W3128235134 cites W1995194116 @default.
• W3128235134 cites W2021347102 @default.
• W3128235134 cites W2108598243 @default.
• W3128235134 cites W2127701282 @default.
• W3128235134 cites W2180612164 @default.
• W3128235134 cites W2194775991 @default.
• W3128235134 cites W2243397390 @default.
• W3128235134 cites W2326925005 @default.
• W3128235134 cites W2408141691 @default.
• W3128235134 cites W2412782625 @default.
• W3128235134 cites W2460937040 @default.
• W3128235134 cites W2519963891 @default.
• W3128235134 cites W2556872594 @default.
• W3128235134 cites W2561498661 @default.
• W3128235134 cites W2603766943 @default.
• W3128235134 cites W2611576673 @default.
• W3128235134 cites W2798302089 @default.
• W3128235134 cites W2798801120 @default.
• W3128235134 cites W2807040120 @default.
• W3128235134 cites W2887906331 @default.
• W3128235134 cites W2905311601 @default.
• W3128235134 cites W2907518016 @default.
• W3128235134 cites W2962360676 @default.
• W3128235134 cites W2962872506 @default.
• W3128235134 cites W2963073614 @default.
• W3128235134 cites W2963143631 @default.
• W3128235134 cites W2963150697 @default.
• W3128235134 cites W2963207607 @default.
• W3128235134 cites W2963361074 @default.
• W3128235134 cites W2963452532 @default.
• W3128235134 cites W2963496101 @default.
• W3128235134 cites W2963767233 @default.
• W3128235134 cites W2963855547 @default.
• W3128235134 cites W2963857521 @default.
• W3128235134 cites W2964082701 @default.
• W3128235134 cites W2964116600 @default.
• W3128235134 cites W2964152294 @default.
• W3128235134 cites W2964153729 @default.
• W3128235134 cites W2964175514 @default.
• W3128235134 cites W2964205597 @default.
• W3128235134 cites W2964253222 @default.
• W3128235134 cites W2964285471 @default.
• W3128235134 cites W2983498217 @default.
• W3128235134 cites W2996603157 @default.
• W3128235134 cites W3034412497 @default.
• W3128235134 cites W3035447895 @default.
• W3128235134 cites W3103557498 @default.
• W3128235134 cites W3105806188 @default.
• W3128235134 cites W3216328231 @default.
• W3128235134 hasPublicationYear "2021" @default.
• W3128235134 type Work @default.
• W3128235134 sameAs 3128235134 @default.
• W3128235134 citedByCount "1" @default.
• W3128235134 countsByYear W31282351342021 @default.
• W3128235134 crossrefType "posted-content" @default.
• W3128235134 hasAuthorship W3128235134A5059313827 @default.
• W3128235134 hasAuthorship W3128235134A5060060011 @default.
• W3128235134 hasAuthorship W3128235134A5085652556 @default.
• W3128235134 hasConcept C104317684 @default.
• W3128235134 hasConcept C108583219 @default.
• W3128235134 hasConcept C119857082 @default.
• W3128235134 hasConcept C127413603 @default.
• W3128235134 hasConcept C153180895 @default.
• W3128235134 hasConcept C154945302 @default.
• W3128235134 hasConcept C175309249 @default.
• W3128235134 hasConcept C185592680 @default.
• W3128235134 hasConcept C31972630 @default.
• W3128235134 hasConcept C3261483 @default.
• W3128235134 hasConcept C37736160 @default.
• W3128235134 hasConcept C41008148 @default.
• W3128235134 hasConcept C50644808 @default.
• W3128235134 hasConcept C55493867 @default.
• W3128235134 hasConcept C63479239 @default.
• W3128235134 hasConcept C65909025 @default.
• W3128235134 hasConcept C76155785 @default.
• W3128235134 hasConcept C81363708 @default.
• W3128235134 hasConcept C82990744 @default.
• W3128235134 hasConcept C87717796 @default.
• W3128235134 hasConcept C94915269 @default.
• W3128235134 hasConcept C95623464 @default.
• W3128235134 hasConceptScore W3128235134C104317684 @default.
• W3128235134 hasConceptScore W3128235134C108583219 @default.
• W3128235134 hasConceptScore W3128235134C119857082 @default.
• W3128235134 hasConceptScore W3128235134C127413603 @default.
• W3128235134 hasConceptScore W3128235134C153180895 @default.
• W3128235134 hasConceptScore W3128235134C154945302 @default.
• W3128235134 hasConceptScore W3128235134C175309249 @default.
• W3128235134 hasConceptScore W3128235134C185592680 @default.
• W3128235134 hasConceptScore W3128235134C31972630 @default.
• W3128235134 hasConceptScore W3128235134C3261483 @default.

• next