Matches in SemOpenAlex for { <https://semopenalex.org/work/W4384663134> ?p ?o ?g. }
Showing items 1 to 71 of
71
with 100 items per page.
- W4384663134 endingPage "9" @default.
- W4384663134 startingPage "1" @default.
- W4384663134 abstract "Adversarial training using empirical risk minimization (ERM) is the state-of-the-art method for defense against adversarial attacks, that is, against small additive adversarial perturbations applied to test data leading to misclassification. Despite being successful in practice, understanding the generalization properties of adversarial training in classification remains widely open. In this article, we take the first step in this direction by precisely characterizing the robustness of adversarial training in binary linear classification. Specifically, we consider the high-dimensional regime where the model dimension grows with the size of the training set at a constant ratio. Our results provide exact asymptotics for both standard and adversarial test errors under general lq -norm bounded perturbations ( q ≥ 1 ) in both discriminative binary models and generative Gaussian-mixture models with correlated features. We use our sharp error formulae to explain how the adversarial and standard errors depend upon the over-parameterization ratio, the data model, and the attack budget. Finally, by comparing with the robust Bayes estimator, our sharp asymptotics allow us to study the fundamental limits of adversarial training." @default.
- W4384663134 created "2023-07-20" @default.
- W4384663134 creator A5011527101 @default.
- W4384663134 creator A5024812488 @default.
- W4384663134 creator A5040270189 @default.
- W4384663134 date "2023-01-01" @default.
- W4384663134 modified "2023-10-15" @default.
- W4384663134 title "Asymptotic Behavior of Adversarial Training in Binary Linear Classification" @default.
- W4384663134 doi "https://doi.org/10.1109/tnnls.2023.3290592" @default.
- W4384663134 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/37463074" @default.
- W4384663134 hasPublicationYear "2023" @default.
- W4384663134 type Work @default.
- W4384663134 citedByCount "0" @default.
- W4384663134 crossrefType "journal-article" @default.
- W4384663134 hasAuthorship W4384663134A5011527101 @default.
- W4384663134 hasAuthorship W4384663134A5024812488 @default.
- W4384663134 hasAuthorship W4384663134A5040270189 @default.
- W4384663134 hasConcept C105795698 @default.
- W4384663134 hasConcept C119857082 @default.
- W4384663134 hasConcept C121332964 @default.
- W4384663134 hasConcept C12267149 @default.
- W4384663134 hasConcept C134306372 @default.
- W4384663134 hasConcept C154945302 @default.
- W4384663134 hasConcept C163716315 @default.
- W4384663134 hasConcept C185429906 @default.
- W4384663134 hasConcept C33923547 @default.
- W4384663134 hasConcept C34388435 @default.
- W4384663134 hasConcept C37736160 @default.
- W4384663134 hasConcept C41008148 @default.
- W4384663134 hasConcept C48372109 @default.
- W4384663134 hasConcept C62520636 @default.
- W4384663134 hasConcept C66905080 @default.
- W4384663134 hasConcept C94375191 @default.
- W4384663134 hasConcept C97931131 @default.
- W4384663134 hasConceptScore W4384663134C105795698 @default.
- W4384663134 hasConceptScore W4384663134C119857082 @default.
- W4384663134 hasConceptScore W4384663134C121332964 @default.
- W4384663134 hasConceptScore W4384663134C12267149 @default.
- W4384663134 hasConceptScore W4384663134C134306372 @default.
- W4384663134 hasConceptScore W4384663134C154945302 @default.
- W4384663134 hasConceptScore W4384663134C163716315 @default.
- W4384663134 hasConceptScore W4384663134C185429906 @default.
- W4384663134 hasConceptScore W4384663134C33923547 @default.
- W4384663134 hasConceptScore W4384663134C34388435 @default.
- W4384663134 hasConceptScore W4384663134C37736160 @default.
- W4384663134 hasConceptScore W4384663134C41008148 @default.
- W4384663134 hasConceptScore W4384663134C48372109 @default.
- W4384663134 hasConceptScore W4384663134C62520636 @default.
- W4384663134 hasConceptScore W4384663134C66905080 @default.
- W4384663134 hasConceptScore W4384663134C94375191 @default.
- W4384663134 hasConceptScore W4384663134C97931131 @default.
- W4384663134 hasFunder F4320306076 @default.
- W4384663134 hasLocation W43846631341 @default.
- W4384663134 hasLocation W43846631342 @default.
- W4384663134 hasOpenAccess W4384663134 @default.
- W4384663134 hasPrimaryLocation W43846631341 @default.
- W4384663134 hasRelatedWork W2036982747 @default.
- W4384663134 hasRelatedWork W2112343299 @default.
- W4384663134 hasRelatedWork W2353457699 @default.
- W4384663134 hasRelatedWork W3033258370 @default.
- W4384663134 hasRelatedWork W3091951680 @default.
- W4384663134 hasRelatedWork W3215867059 @default.
- W4384663134 hasRelatedWork W4280496914 @default.
- W4384663134 hasRelatedWork W4322008322 @default.
- W4384663134 hasRelatedWork W4379255972 @default.
- W4384663134 hasRelatedWork W4383955378 @default.
- W4384663134 isParatext "false" @default.
- W4384663134 isRetracted "false" @default.
- W4384663134 workType "article" @default.