Matches in SemOpenAlex for { <https://semopenalex.org/work/W2950381305> ?p ?o ?g. }
Showing items 1 to 95 of
95
with 100 items per page.
- W2950381305 abstract "Deep neural networks (DNNs) have transformed several artificial intelligence research areas including computer vision, speech recognition, and natural language processing. However, recent studies demonstrated that DNNs are vulnerable to adversarial manipulations at testing time. Specifically, suppose we have a testing example, whose label can be correctly predicted by a DNN classifier. An attacker can add a small carefully crafted noise to the testing example such that the DNN classifier predicts an incorrect label, where the crafted testing example is called adversarial example. Such attacks are called evasion attacks. Evasion attacks are one of the biggest challenges for deploying DNNs in safety and security critical applications such as self-driving cars. In this work, we develop new methods to defend against evasion attacks. Our key observation is that adversarial examples are close to the classification boundary. Therefore, we propose region-based classification to be robust to adversarial examples. For a benign/adversarial testing example, we ensemble information in a hypercube centered at the example to predict its label. In contrast, traditional classifiers are point-based classification, i.e., given a testing example, the classifier predicts its label based on the testing example alone. Our evaluation results on MNIST and CIFAR-10 datasets demonstrate that our region-based classification can significantly mitigate evasion attacks without sacrificing classification accuracy on benign examples. Specifically, our region-based classification achieves the same classification accuracy on testing benign examples as point-based classification, but our region-based classification is significantly more robust than point-based classification to various evasion attacks." @default.
- W2950381305 created "2019-06-27" @default.
- W2950381305 creator A5009102659 @default.
- W2950381305 creator A5032504910 @default.
- W2950381305 date "2017-09-16" @default.
- W2950381305 modified "2023-09-26" @default.
- W2950381305 title "Mitigating Evasion Attacks to Deep Neural Networks via Region-based Classification" @default.
- W2950381305 cites W1522301498 @default.
- W2950381305 cites W1673923490 @default.
- W2950381305 cites W1945616565 @default.
- W2950381305 cites W2095577883 @default.
- W2950381305 cites W2140793251 @default.
- W2950381305 cites W2162552722 @default.
- W2950381305 cites W2194775991 @default.
- W2950381305 cites W2243397390 @default.
- W2950381305 cites W2257979135 @default.
- W2950381305 cites W2460937040 @default.
- W2950381305 cites W2559840118 @default.
- W2950381305 cites W2561975083 @default.
- W2950381305 cites W2590523583 @default.
- W2950381305 cites W2593892853 @default.
- W2950381305 cites W2594867206 @default.
- W2950381305 cites W2605631833 @default.
- W2950381305 cites W2607219512 @default.
- W2950381305 cites W2610321374 @default.
- W2950381305 cites W2618043096 @default.
- W2950381305 cites W2619479788 @default.
- W2950381305 cites W2640329709 @default.
- W2950381305 cites W2767075075 @default.
- W2950381305 cites W2950577311 @default.
- W2950381305 cites W2950864148 @default.
- W2950381305 cites W2963154688 @default.
- W2950381305 cites W2963389226 @default.
- W2950381305 cites W2963695663 @default.
- W2950381305 cites W2963744840 @default.
- W2950381305 cites W2963857521 @default.
- W2950381305 cites W2964082701 @default.
- W2950381305 doi "https://doi.org/10.48550/arxiv.1709.05583" @default.
- W2950381305 hasPublicationYear "2017" @default.
- W2950381305 type Work @default.
- W2950381305 sameAs 2950381305 @default.
- W2950381305 citedByCount "3" @default.
- W2950381305 countsByYear W29503813052018 @default.
- W2950381305 countsByYear W29503813052019 @default.
- W2950381305 crossrefType "posted-content" @default.
- W2950381305 hasAuthorship W2950381305A5009102659 @default.
- W2950381305 hasAuthorship W2950381305A5032504910 @default.
- W2950381305 hasBestOaLocation W29503813051 @default.
- W2950381305 hasConcept C115961682 @default.
- W2950381305 hasConcept C119857082 @default.
- W2950381305 hasConcept C153180895 @default.
- W2950381305 hasConcept C154945302 @default.
- W2950381305 hasConcept C190502265 @default.
- W2950381305 hasConcept C203014093 @default.
- W2950381305 hasConcept C2781251061 @default.
- W2950381305 hasConcept C2984842247 @default.
- W2950381305 hasConcept C37736160 @default.
- W2950381305 hasConcept C41008148 @default.
- W2950381305 hasConcept C50644808 @default.
- W2950381305 hasConcept C75294576 @default.
- W2950381305 hasConcept C86803240 @default.
- W2950381305 hasConcept C8891405 @default.
- W2950381305 hasConcept C95623464 @default.
- W2950381305 hasConceptScore W2950381305C115961682 @default.
- W2950381305 hasConceptScore W2950381305C119857082 @default.
- W2950381305 hasConceptScore W2950381305C153180895 @default.
- W2950381305 hasConceptScore W2950381305C154945302 @default.
- W2950381305 hasConceptScore W2950381305C190502265 @default.
- W2950381305 hasConceptScore W2950381305C203014093 @default.
- W2950381305 hasConceptScore W2950381305C2781251061 @default.
- W2950381305 hasConceptScore W2950381305C2984842247 @default.
- W2950381305 hasConceptScore W2950381305C37736160 @default.
- W2950381305 hasConceptScore W2950381305C41008148 @default.
- W2950381305 hasConceptScore W2950381305C50644808 @default.
- W2950381305 hasConceptScore W2950381305C75294576 @default.
- W2950381305 hasConceptScore W2950381305C86803240 @default.
- W2950381305 hasConceptScore W2950381305C8891405 @default.
- W2950381305 hasConceptScore W2950381305C95623464 @default.
- W2950381305 hasLocation W29503813051 @default.
- W2950381305 hasOpenAccess W2950381305 @default.
- W2950381305 hasPrimaryLocation W29503813051 @default.
- W2950381305 hasRelatedWork W2604147826 @default.
- W2950381305 hasRelatedWork W2950381305 @default.
- W2950381305 hasRelatedWork W2972262177 @default.
- W2950381305 hasRelatedWork W2972781078 @default.
- W2950381305 hasRelatedWork W2986674980 @default.
- W2950381305 hasRelatedWork W3026616975 @default.
- W2950381305 hasRelatedWork W3109062656 @default.
- W2950381305 hasRelatedWork W4221015625 @default.
- W2950381305 hasRelatedWork W4293054861 @default.
- W2950381305 hasRelatedWork W4297776111 @default.
- W2950381305 isParatext "false" @default.
- W2950381305 isRetracted "false" @default.
- W2950381305 magId "2950381305" @default.
- W2950381305 workType "article" @default.