Matches in SemOpenAlex for { <https://semopenalex.org/work/W3122142687> ?p ?o ?g. }
Showing items 1 to 87 of
87
with 100 items per page.
- W3122142687 abstract "Learning a distance metric between pairs of examples is widely important for various tasks. Deep Metric Learning (DML) utilizes deep neural network architectures to learn semantic feature embeddings where the distance between similar examples is close and dissimilar examples are far. While the underlying neural networks produce good accuracy on naturally occurring samples, they are vulnerable to adversarially-perturbed samples that can reduce their accuracy. To create robust versions of DML models, we introduce a robust training approach. A key challenge is that metric losses are not independent --- they depend on all samples in a mini-batch. This sensitivity to samples, if not accounted for, can lead to incorrect robust training. To the best of our knowledge, we are the first to systematically analyze this dependence effect and propose a principled approach for robust training of deep metric learning networks that accounts for the nuances of metric losses. Using experiments on three popular datasets in metric learning, we demonstrate the DML models trained using our techniques display robustness against strong iterative attacks while their performance on unperturbed (natural) samples remains largely unaffected." @default.
- W3122142687 created "2021-02-01" @default.
- W3122142687 creator A5013838615 @default.
- W3122142687 creator A5051503339 @default.
- W3122142687 creator A5060924315 @default.
- W3122142687 creator A5078821760 @default.
- W3122142687 creator A5088826068 @default.
- W3122142687 date "2021-05-04" @default.
- W3122142687 modified "2023-10-13" @default.
- W3122142687 title "Adversarial Deep Metric Learning" @default.
- W3122142687 hasPublicationYear "2021" @default.
- W3122142687 type Work @default.
- W3122142687 sameAs 3122142687 @default.
- W3122142687 citedByCount "0" @default.
- W3122142687 crossrefType "journal-article" @default.
- W3122142687 hasAuthorship W3122142687A5013838615 @default.
- W3122142687 hasAuthorship W3122142687A5051503339 @default.
- W3122142687 hasAuthorship W3122142687A5060924315 @default.
- W3122142687 hasAuthorship W3122142687A5078821760 @default.
- W3122142687 hasAuthorship W3122142687A5088826068 @default.
- W3122142687 hasConcept C104317684 @default.
- W3122142687 hasConcept C108583219 @default.
- W3122142687 hasConcept C119857082 @default.
- W3122142687 hasConcept C127413603 @default.
- W3122142687 hasConcept C138885662 @default.
- W3122142687 hasConcept C153180895 @default.
- W3122142687 hasConcept C154945302 @default.
- W3122142687 hasConcept C176217482 @default.
- W3122142687 hasConcept C185592680 @default.
- W3122142687 hasConcept C21547014 @default.
- W3122142687 hasConcept C26517878 @default.
- W3122142687 hasConcept C2776401178 @default.
- W3122142687 hasConcept C2984842247 @default.
- W3122142687 hasConcept C37736160 @default.
- W3122142687 hasConcept C38652104 @default.
- W3122142687 hasConcept C41008148 @default.
- W3122142687 hasConcept C41895202 @default.
- W3122142687 hasConcept C50644808 @default.
- W3122142687 hasConcept C55493867 @default.
- W3122142687 hasConcept C63479239 @default.
- W3122142687 hasConceptScore W3122142687C104317684 @default.
- W3122142687 hasConceptScore W3122142687C108583219 @default.
- W3122142687 hasConceptScore W3122142687C119857082 @default.
- W3122142687 hasConceptScore W3122142687C127413603 @default.
- W3122142687 hasConceptScore W3122142687C138885662 @default.
- W3122142687 hasConceptScore W3122142687C153180895 @default.
- W3122142687 hasConceptScore W3122142687C154945302 @default.
- W3122142687 hasConceptScore W3122142687C176217482 @default.
- W3122142687 hasConceptScore W3122142687C185592680 @default.
- W3122142687 hasConceptScore W3122142687C21547014 @default.
- W3122142687 hasConceptScore W3122142687C26517878 @default.
- W3122142687 hasConceptScore W3122142687C2776401178 @default.
- W3122142687 hasConceptScore W3122142687C2984842247 @default.
- W3122142687 hasConceptScore W3122142687C37736160 @default.
- W3122142687 hasConceptScore W3122142687C38652104 @default.
- W3122142687 hasConceptScore W3122142687C41008148 @default.
- W3122142687 hasConceptScore W3122142687C41895202 @default.
- W3122142687 hasConceptScore W3122142687C50644808 @default.
- W3122142687 hasConceptScore W3122142687C55493867 @default.
- W3122142687 hasConceptScore W3122142687C63479239 @default.
- W3122142687 hasLocation W31221426871 @default.
- W3122142687 hasOpenAccess W3122142687 @default.
- W3122142687 hasPrimaryLocation W31221426871 @default.
- W3122142687 hasRelatedWork W2245869277 @default.
- W3122142687 hasRelatedWork W2781758978 @default.
- W3122142687 hasRelatedWork W2787637945 @default.
- W3122142687 hasRelatedWork W2884169041 @default.
- W3122142687 hasRelatedWork W2905582115 @default.
- W3122142687 hasRelatedWork W2913505554 @default.
- W3122142687 hasRelatedWork W2963501948 @default.
- W3122142687 hasRelatedWork W2963753390 @default.
- W3122142687 hasRelatedWork W2981681681 @default.
- W3122142687 hasRelatedWork W2995645057 @default.
- W3122142687 hasRelatedWork W2996687884 @default.
- W3122142687 hasRelatedWork W3008051686 @default.
- W3122142687 hasRelatedWork W3018911559 @default.
- W3122142687 hasRelatedWork W3028525609 @default.
- W3122142687 hasRelatedWork W3033258370 @default.
- W3122142687 hasRelatedWork W3034349555 @default.
- W3122142687 hasRelatedWork W3042543350 @default.
- W3122142687 hasRelatedWork W3093152201 @default.
- W3122142687 hasRelatedWork W3104032928 @default.
- W3122142687 hasRelatedWork W3161366266 @default.
- W3122142687 isParatext "false" @default.
- W3122142687 isRetracted "false" @default.
- W3122142687 magId "3122142687" @default.
- W3122142687 workType "article" @default.