Matches in SemOpenAlex for { <https://semopenalex.org/work/W4312457673> ?p ?o ?g. }
Showing items 1 to 98 of
98
with 100 items per page.
- W4312457673 abstract "Deep learning based techniques are broadly used in a variety of applications such as image recognition, natural language processing, etc., which express leading performance than traditional methods. However, adversarial samples can cause severe problems for deep learning models, especially for vision-based deep neural networks. Adversarial training is a commonly adopted strategy to defend adversarial samples, but it lacks generalization capability due to the transferability of adversarial samples to work on cross-task predictions. In addition, the input processing method of deep learning pipeline can improve generalization capability for cross-task predictions, whereas processing newly generated adversarial samples is the major bottleneck that limits its performance. To deal with these issues, this paper proposes a purification mechanism that combines both adversarial training-based and input processing-based methods to defend the threats of adversarial samples. The proposed mechanism has a strong generalization capability for various vision-based deep learning tasks by leveraging the advantage of input processing-based methods and maintaining the defense efficiency of adversarial training-based methods. Comprehensive experiments are conducted on multiple datasets that demonstrate the superiority of the proposed approach against the state-of-the-art methods on defending adversarial samples." @default.
- W4312457673 created "2023-01-04" @default.
- W4312457673 creator A5032891611 @default.
- W4312457673 creator A5034199009 @default.
- W4312457673 creator A5046635673 @default.
- W4312457673 creator A5054098183 @default.
- W4312457673 creator A5064192743 @default.
- W4312457673 date "2022-08-01" @default.
- W4312457673 modified "2023-09-24" @default.
- W4312457673 title "A Self-Supervised Purification Mechanism for Adversarial Samples" @default.
- W4312457673 cites W2064675550 @default.
- W4312457673 cites W2139137304 @default.
- W4312457673 cites W2325939864 @default.
- W4312457673 cites W2401520370 @default.
- W4312457673 cites W2533800772 @default.
- W4312457673 cites W2604249033 @default.
- W4312457673 cites W2772693146 @default.
- W4312457673 cites W2774018344 @default.
- W4312457673 cites W2794284562 @default.
- W4312457673 cites W2806031239 @default.
- W4312457673 cites W2884001105 @default.
- W4312457673 cites W2905810301 @default.
- W4312457673 cites W2963150697 @default.
- W4312457673 cites W2963178695 @default.
- W4312457673 cites W2963448658 @default.
- W4312457673 cites W2969542116 @default.
- W4312457673 cites W2969664989 @default.
- W4312457673 cites W3019166713 @default.
- W4312457673 cites W3023402713 @default.
- W4312457673 cites W3034190247 @default.
- W4312457673 cites W3035078899 @default.
- W4312457673 cites W3043149731 @default.
- W4312457673 cites W3118836214 @default.
- W4312457673 cites W3170019900 @default.
- W4312457673 cites W3177624059 @default.
- W4312457673 doi "https://doi.org/10.1109/ithings-greencom-cpscom-smartdata-cybermatics55523.2022.00102" @default.
- W4312457673 hasPublicationYear "2022" @default.
- W4312457673 type Work @default.
- W4312457673 citedByCount "0" @default.
- W4312457673 crossrefType "proceedings-article" @default.
- W4312457673 hasAuthorship W4312457673A5032891611 @default.
- W4312457673 hasAuthorship W4312457673A5034199009 @default.
- W4312457673 hasAuthorship W4312457673A5046635673 @default.
- W4312457673 hasAuthorship W4312457673A5054098183 @default.
- W4312457673 hasAuthorship W4312457673A5064192743 @default.
- W4312457673 hasConcept C108583219 @default.
- W4312457673 hasConcept C119857082 @default.
- W4312457673 hasConcept C127413603 @default.
- W4312457673 hasConcept C134306372 @default.
- W4312457673 hasConcept C149635348 @default.
- W4312457673 hasConcept C154945302 @default.
- W4312457673 hasConcept C177148314 @default.
- W4312457673 hasConcept C199360897 @default.
- W4312457673 hasConcept C201995342 @default.
- W4312457673 hasConcept C2780451532 @default.
- W4312457673 hasConcept C2780513914 @default.
- W4312457673 hasConcept C2984842247 @default.
- W4312457673 hasConcept C33923547 @default.
- W4312457673 hasConcept C37736160 @default.
- W4312457673 hasConcept C41008148 @default.
- W4312457673 hasConcept C43521106 @default.
- W4312457673 hasConcept C50644808 @default.
- W4312457673 hasConcept C94124525 @default.
- W4312457673 hasConceptScore W4312457673C108583219 @default.
- W4312457673 hasConceptScore W4312457673C119857082 @default.
- W4312457673 hasConceptScore W4312457673C127413603 @default.
- W4312457673 hasConceptScore W4312457673C134306372 @default.
- W4312457673 hasConceptScore W4312457673C149635348 @default.
- W4312457673 hasConceptScore W4312457673C154945302 @default.
- W4312457673 hasConceptScore W4312457673C177148314 @default.
- W4312457673 hasConceptScore W4312457673C199360897 @default.
- W4312457673 hasConceptScore W4312457673C201995342 @default.
- W4312457673 hasConceptScore W4312457673C2780451532 @default.
- W4312457673 hasConceptScore W4312457673C2780513914 @default.
- W4312457673 hasConceptScore W4312457673C2984842247 @default.
- W4312457673 hasConceptScore W4312457673C33923547 @default.
- W4312457673 hasConceptScore W4312457673C37736160 @default.
- W4312457673 hasConceptScore W4312457673C41008148 @default.
- W4312457673 hasConceptScore W4312457673C43521106 @default.
- W4312457673 hasConceptScore W4312457673C50644808 @default.
- W4312457673 hasConceptScore W4312457673C94124525 @default.
- W4312457673 hasFunder F4320306076 @default.
- W4312457673 hasLocation W43124576731 @default.
- W4312457673 hasOpenAccess W4312457673 @default.
- W4312457673 hasPrimaryLocation W43124576731 @default.
- W4312457673 hasRelatedWork W2949152835 @default.
- W4312457673 hasRelatedWork W2950066684 @default.
- W4312457673 hasRelatedWork W2952919291 @default.
- W4312457673 hasRelatedWork W3082895349 @default.
- W4312457673 hasRelatedWork W3193857078 @default.
- W4312457673 hasRelatedWork W4293054861 @default.
- W4312457673 hasRelatedWork W4298388782 @default.
- W4312457673 hasRelatedWork W4300837091 @default.
- W4312457673 hasRelatedWork W4312831135 @default.
- W4312457673 hasRelatedWork W4320076403 @default.
- W4312457673 isParatext "false" @default.
- W4312457673 isRetracted "false" @default.
- W4312457673 workType "article" @default.