Matches in SemOpenAlex for { <https://semopenalex.org/work/W4285298979> ?p ?o ?g. }
- W4285298979 endingPage "196" @default.
- W4285298979 startingPage "171" @default.
- W4285298979 abstract "Abstract Although deep neural networks (DNNs) are high-performance methods for various complex tasks, e.g., environment perception in automated vehicles (AVs), they are vulnerable to adversarial perturbations. Recent works have proven the existence of universal adversarial perturbations (UAPs), which, when added to most images, destroy the output of the respective perception function. Existing attack methods often show a low success rate when attacking target models which are different from the one that the attack was optimized on. To address such weak transferability, we propose a novel learning criterion by combining a low-level feature loss, addressing the similarity of feature representations in the first layer of various model architectures, with a cross-entropy loss. Experimental results on ImageNet and Cityscapes datasets show that our method effectively generates universal adversarial perturbations achieving state-of-the-art fooling rates across different models, tasks, and datasets. Due to their effectiveness, we propose the use of such novel generated UAPs in robustness evaluation of DNN-based environment perception functions for AVs." @default.
- W4285298979 created "2022-07-14" @default.
- W4285298979 creator A5002593702 @default.
- W4285298979 creator A5018314455 @default.
- W4285298979 creator A5029540135 @default.
- W4285298979 creator A5087966698 @default.
- W4285298979 date "2022-01-01" @default.
- W4285298979 modified "2023-09-26" @default.
- W4285298979 title "Improving Transferability of Generated Universal Adversarial Perturbations for Image Classification and Segmentation" @default.
- W4285298979 cites W1903029394 @default.
- W4285298979 cites W2064076387 @default.
- W4285298979 cites W2097117768 @default.
- W4285298979 cites W2117539524 @default.
- W4285298979 cites W2133665775 @default.
- W4285298979 cites W2180612164 @default.
- W4285298979 cites W2194775991 @default.
- W4285298979 cites W2331128040 @default.
- W4285298979 cites W2340897893 @default.
- W4285298979 cites W2543927648 @default.
- W4285298979 cites W2592680288 @default.
- W4285298979 cites W2762439315 @default.
- W4285298979 cites W2774644650 @default.
- W4285298979 cites W2798412221 @default.
- W4285298979 cites W2885039223 @default.
- W4285298979 cites W2898988942 @default.
- W4285298979 cites W2949789602 @default.
- W4285298979 cites W2955364291 @default.
- W4285298979 cites W2962700793 @default.
- W4285298979 cites W2962793481 @default.
- W4285298979 cites W2963118571 @default.
- W4285298979 cites W2963165363 @default.
- W4285298979 cites W2963560523 @default.
- W4285298979 cites W2963697717 @default.
- W4285298979 cites W2963855547 @default.
- W4285298979 cites W2964097310 @default.
- W4285298979 cites W2964171870 @default.
- W4285298979 cites W2965198951 @default.
- W4285298979 cites W2979491204 @default.
- W4285298979 cites W2981207549 @default.
- W4285298979 cites W2985495028 @default.
- W4285298979 cites W2997583194 @default.
- W4285298979 cites W2997868396 @default.
- W4285298979 cites W2998421476 @default.
- W4285298979 cites W3011074071 @default.
- W4285298979 cites W3034464529 @default.
- W4285298979 cites W3034643863 @default.
- W4285298979 cites W3035004134 @default.
- W4285298979 cites W3045897451 @default.
- W4285298979 cites W3106835084 @default.
- W4285298979 cites W3108534877 @default.
- W4285298979 cites W3117786822 @default.
- W4285298979 doi "https://doi.org/10.1007/978-3-031-01233-4_6" @default.
- W4285298979 hasPublicationYear "2022" @default.
- W4285298979 type Work @default.
- W4285298979 citedByCount "3" @default.
- W4285298979 countsByYear W42852989792022 @default.
- W4285298979 countsByYear W42852989792023 @default.
- W4285298979 crossrefType "book-chapter" @default.
- W4285298979 hasAuthorship W4285298979A5002593702 @default.
- W4285298979 hasAuthorship W4285298979A5018314455 @default.
- W4285298979 hasAuthorship W4285298979A5029540135 @default.
- W4285298979 hasAuthorship W4285298979A5087966698 @default.
- W4285298979 hasBestOaLocation W42852989791 @default.
- W4285298979 hasConcept C104317684 @default.
- W4285298979 hasConcept C108583219 @default.
- W4285298979 hasConcept C119857082 @default.
- W4285298979 hasConcept C138885662 @default.
- W4285298979 hasConcept C140331021 @default.
- W4285298979 hasConcept C153180895 @default.
- W4285298979 hasConcept C154945302 @default.
- W4285298979 hasConcept C169760540 @default.
- W4285298979 hasConcept C185592680 @default.
- W4285298979 hasConcept C26760741 @default.
- W4285298979 hasConcept C2776401178 @default.
- W4285298979 hasConcept C2984842247 @default.
- W4285298979 hasConcept C37736160 @default.
- W4285298979 hasConcept C41008148 @default.
- W4285298979 hasConcept C41895202 @default.
- W4285298979 hasConcept C55493867 @default.
- W4285298979 hasConcept C61272859 @default.
- W4285298979 hasConcept C63479239 @default.
- W4285298979 hasConcept C86803240 @default.
- W4285298979 hasConcept C89600930 @default.
- W4285298979 hasConceptScore W4285298979C104317684 @default.
- W4285298979 hasConceptScore W4285298979C108583219 @default.
- W4285298979 hasConceptScore W4285298979C119857082 @default.
- W4285298979 hasConceptScore W4285298979C138885662 @default.
- W4285298979 hasConceptScore W4285298979C140331021 @default.
- W4285298979 hasConceptScore W4285298979C153180895 @default.
- W4285298979 hasConceptScore W4285298979C154945302 @default.
- W4285298979 hasConceptScore W4285298979C169760540 @default.
- W4285298979 hasConceptScore W4285298979C185592680 @default.
- W4285298979 hasConceptScore W4285298979C26760741 @default.
- W4285298979 hasConceptScore W4285298979C2776401178 @default.
- W4285298979 hasConceptScore W4285298979C2984842247 @default.
- W4285298979 hasConceptScore W4285298979C37736160 @default.
- W4285298979 hasConceptScore W4285298979C41008148 @default.
- W4285298979 hasConceptScore W4285298979C41895202 @default.