Matches in SemOpenAlex for { <https://semopenalex.org/work/W3043789969> ?p ?o ?g. }
- W3043789969 abstract "Deep neural networks (DNNs) have been proven vulnerable to backdoor attacks, where hidden features (patterns) trained to a normal model, which is only activated by some specific input (called triggers), trick the model into producing unexpected behavior. In this paper, we create covert and scattered triggers for backdoor attacks, invisible backdoors, where triggers can fool both DNN models and human inspection. We apply our invisible backdoors through two state-of-the-art methods of embedding triggers for backdoor attacks. The first approach on Badnets embeds the trigger into DNNs through steganography. The second approach of a trojan attack uses two types of additional regularization terms to generate the triggers with irregular shape and size. We use the Attack Success Rate and Functionality to measure the performance of our attacks. We introduce two novel definitions of invisibility for human perception; one is conceptualized by the Perceptual Adversarial Similarity Score (PASS) and the other is Learned Perceptual Image Patch Similarity (LPIPS). We show that the proposed invisible backdoors can be fairly effective across various DNN models as well as four datasets MNIST, CIFAR-10, CIFAR-100, and GTSRB, by measuring their attack success rates for the adversary, functionality for the normal users, and invisibility scores for the administrators. We finally argue that the proposed invisible backdoor attacks can effectively thwart the state-of-the-art trojan backdoor detection approaches, such as Neural Cleanse and TABOR." @default.
- W3043789969 created "2020-07-23" @default.
- W3043789969 creator A5009850797 @default.
- W3043789969 creator A5016670591 @default.
- W3043789969 creator A5016909512 @default.
- W3043789969 creator A5041505236 @default.
- W3043789969 creator A5071724015 @default.
- W3043789969 date "2019-09-06" @default.
- W3043789969 modified "2023-10-07" @default.
- W3043789969 title "Invisible Backdoor Attacks on Deep Neural Networks via Steganography and Regularization" @default.
- W3043789969 cites W1524144700 @default.
- W3043789969 cites W1806317803 @default.
- W3043789969 cites W2002427601 @default.
- W3043789969 cites W2067713319 @default.
- W3043789969 cites W2145339207 @default.
- W3043789969 cites W2153631546 @default.
- W3043789969 cites W2160815625 @default.
- W3043789969 cites W2163922914 @default.
- W3043789969 cites W2167421362 @default.
- W3043789969 cites W2180612164 @default.
- W3043789969 cites W2194775991 @default.
- W3043789969 cites W2257979135 @default.
- W3043789969 cites W2473418344 @default.
- W3043789969 cites W2535690855 @default.
- W3043789969 cites W2543927648 @default.
- W3043789969 cites W2557044351 @default.
- W3043789969 cites W2571946726 @default.
- W3043789969 cites W2619479788 @default.
- W3043789969 cites W2748789698 @default.
- W3043789969 cites W2753783305 @default.
- W3043789969 cites W2772825438 @default.
- W3043789969 cites W2773446523 @default.
- W3043789969 cites W2807363941 @default.
- W3043789969 cites W2934843808 @default.
- W3043789969 cites W2937920463 @default.
- W3043789969 cites W2942091739 @default.
- W3043789969 cites W2942630857 @default.
- W3043789969 cites W2962748759 @default.
- W3043789969 cites W2962785568 @default.
- W3043789969 cites W2963098487 @default.
- W3043789969 cites W2963177963 @default.
- W3043789969 cites W2963207607 @default.
- W3043789969 cites W2963857521 @default.
- W3043789969 cites W2963888996 @default.
- W3043789969 cites W2963952467 @default.
- W3043789969 cites W2964153729 @default.
- W3043789969 cites W2964171870 @default.
- W3043789969 cites W2964253222 @default.
- W3043789969 cites W2965527544 @default.
- W3043789969 cites W2966187620 @default.
- W3043789969 cites W2967540978 @default.
- W3043789969 cites W2969542116 @default.
- W3043789969 cites W2986013765 @default.
- W3043789969 cites W2990270730 @default.
- W3043789969 cites W2997502936 @default.
- W3043789969 cites W3015716673 @default.
- W3043789969 cites W3088733693 @default.
- W3043789969 cites W3118608800 @default.
- W3043789969 cites W34239198 @default.
- W3043789969 cites W2019048242 @default.
- W3043789969 cites W2470534717 @default.
- W3043789969 doi "https://doi.org/10.48550/arxiv.1909.02742" @default.
- W3043789969 hasPublicationYear "2019" @default.
- W3043789969 type Work @default.
- W3043789969 sameAs 3043789969 @default.
- W3043789969 citedByCount "14" @default.
- W3043789969 countsByYear W30437899692018 @default.
- W3043789969 countsByYear W30437899692020 @default.
- W3043789969 countsByYear W30437899692021 @default.
- W3043789969 countsByYear W30437899692022 @default.
- W3043789969 crossrefType "posted-content" @default.
- W3043789969 hasAuthorship W3043789969A5009850797 @default.
- W3043789969 hasAuthorship W3043789969A5016670591 @default.
- W3043789969 hasAuthorship W3043789969A5016909512 @default.
- W3043789969 hasAuthorship W3043789969A5041505236 @default.
- W3043789969 hasAuthorship W3043789969A5071724015 @default.
- W3043789969 hasBestOaLocation W30437899691 @default.
- W3043789969 hasConcept C108583219 @default.
- W3043789969 hasConcept C119857082 @default.
- W3043789969 hasConcept C153180895 @default.
- W3043789969 hasConcept C154945302 @default.
- W3043789969 hasConcept C15744967 @default.
- W3043789969 hasConcept C169760540 @default.
- W3043789969 hasConcept C174333608 @default.
- W3043789969 hasConcept C26760741 @default.
- W3043789969 hasConcept C2781045450 @default.
- W3043789969 hasConcept C2984842247 @default.
- W3043789969 hasConcept C38652104 @default.
- W3043789969 hasConcept C41008148 @default.
- W3043789969 hasConcept C41065033 @default.
- W3043789969 hasConcept C41608201 @default.
- W3043789969 hasConcept C50962388 @default.
- W3043789969 hasConceptScore W3043789969C108583219 @default.
- W3043789969 hasConceptScore W3043789969C119857082 @default.
- W3043789969 hasConceptScore W3043789969C153180895 @default.
- W3043789969 hasConceptScore W3043789969C154945302 @default.
- W3043789969 hasConceptScore W3043789969C15744967 @default.
- W3043789969 hasConceptScore W3043789969C169760540 @default.
- W3043789969 hasConceptScore W3043789969C174333608 @default.
- W3043789969 hasConceptScore W3043789969C26760741 @default.