Matches in SemOpenAlex for { <https://semopenalex.org/work/W2912272583> ?p ?o ?g. }
Showing items 1 to 97 of
97
with 100 items per page.
- W2912272583 abstract "Recently, there has been an abundance of works on designing Deep Neural Networks (DNNs) that are robust to adversarial examples. In particular, a central question is which features of DNNs influence adversarial robustness and, therefore, can be to used to design robust DNNs. In this work, this problem is studied through the lens of compression which is captured by the low-rank structure of weight matrices. It is first shown that adversarial training tends to promote simultaneously low-rank and sparse structure in the weight matrices of neural networks. This is measured through the notions of effective rank and effective sparsity. In the reverse direction, when the low rank structure is promoted by nuclear norm regularization and combined with sparsity inducing regularizations, neural networks show significantly improved adversarial robustness. The effect of nuclear norm regularization on adversarial robustness is paramount when it is applied to convolutional neural networks. Although still not competing with adversarial training, this result contributes to understanding the key properties of robust classifiers." @default.
- W2912272583 created "2019-02-21" @default.
- W2912272583 creator A5023739887 @default.
- W2912272583 creator A5066277118 @default.
- W2912272583 creator A5067916678 @default.
- W2912272583 creator A5083706779 @default.
- W2912272583 date "2019-01-29" @default.
- W2912272583 modified "2023-09-27" @default.
- W2912272583 title "On the Effect of Low-Rank Weights on Adversarial Robustness of Neural Networks." @default.
- W2912272583 cites W1677182931 @default.
- W2912272583 cites W182164380 @default.
- W2912272583 cites W2296673577 @default.
- W2912272583 cites W2513314332 @default.
- W2912272583 cites W2533641151 @default.
- W2912272583 cites W2543927648 @default.
- W2912272583 cites W2552767274 @default.
- W2912272583 cites W2558893580 @default.
- W2912272583 cites W2592929672 @default.
- W2912272583 cites W2593634001 @default.
- W2912272583 cites W2612637113 @default.
- W2912272583 cites W2613509922 @default.
- W2912272583 cites W2759071281 @default.
- W2912272583 cites W2784285928 @default.
- W2912272583 cites W2785887350 @default.
- W2912272583 cites W2790361456 @default.
- W2912272583 cites W2797081606 @default.
- W2912272583 cites W2891710009 @default.
- W2912272583 cites W2963467071 @default.
- W2912272583 cites W2963955657 @default.
- W2912272583 cites W2964043206 @default.
- W2912272583 cites W2964184826 @default.
- W2912272583 cites W2964248288 @default.
- W2912272583 cites W3106022083 @default.
- W2912272583 hasPublicationYear "2019" @default.
- W2912272583 type Work @default.
- W2912272583 sameAs 2912272583 @default.
- W2912272583 citedByCount "5" @default.
- W2912272583 countsByYear W29122725832020 @default.
- W2912272583 countsByYear W29122725832021 @default.
- W2912272583 crossrefType "posted-content" @default.
- W2912272583 hasAuthorship W2912272583A5023739887 @default.
- W2912272583 hasAuthorship W2912272583A5066277118 @default.
- W2912272583 hasAuthorship W2912272583A5067916678 @default.
- W2912272583 hasAuthorship W2912272583A5083706779 @default.
- W2912272583 hasConcept C104317684 @default.
- W2912272583 hasConcept C11413529 @default.
- W2912272583 hasConcept C119857082 @default.
- W2912272583 hasConcept C154945302 @default.
- W2912272583 hasConcept C185592680 @default.
- W2912272583 hasConcept C2776135515 @default.
- W2912272583 hasConcept C2984842247 @default.
- W2912272583 hasConcept C37736160 @default.
- W2912272583 hasConcept C41008148 @default.
- W2912272583 hasConcept C50644808 @default.
- W2912272583 hasConcept C55493867 @default.
- W2912272583 hasConcept C63479239 @default.
- W2912272583 hasConcept C81363708 @default.
- W2912272583 hasConceptScore W2912272583C104317684 @default.
- W2912272583 hasConceptScore W2912272583C11413529 @default.
- W2912272583 hasConceptScore W2912272583C119857082 @default.
- W2912272583 hasConceptScore W2912272583C154945302 @default.
- W2912272583 hasConceptScore W2912272583C185592680 @default.
- W2912272583 hasConceptScore W2912272583C2776135515 @default.
- W2912272583 hasConceptScore W2912272583C2984842247 @default.
- W2912272583 hasConceptScore W2912272583C37736160 @default.
- W2912272583 hasConceptScore W2912272583C41008148 @default.
- W2912272583 hasConceptScore W2912272583C50644808 @default.
- W2912272583 hasConceptScore W2912272583C55493867 @default.
- W2912272583 hasConceptScore W2912272583C63479239 @default.
- W2912272583 hasConceptScore W2912272583C81363708 @default.
- W2912272583 hasLocation W29122725831 @default.
- W2912272583 hasOpenAccess W2912272583 @default.
- W2912272583 hasPrimaryLocation W29122725831 @default.
- W2912272583 hasRelatedWork W1883420340 @default.
- W2912272583 hasRelatedWork W2180290696 @default.
- W2912272583 hasRelatedWork W2619682061 @default.
- W2912272583 hasRelatedWork W2900578291 @default.
- W2912272583 hasRelatedWork W2947600649 @default.
- W2912272583 hasRelatedWork W2947622387 @default.
- W2912272583 hasRelatedWork W2980749867 @default.
- W2912272583 hasRelatedWork W3036835756 @default.
- W2912272583 hasRelatedWork W3083717542 @default.
- W2912272583 hasRelatedWork W3089485619 @default.
- W2912272583 hasRelatedWork W3115334024 @default.
- W2912272583 hasRelatedWork W3125845062 @default.
- W2912272583 hasRelatedWork W3131236472 @default.
- W2912272583 hasRelatedWork W3160981463 @default.
- W2912272583 hasRelatedWork W3165462161 @default.
- W2912272583 hasRelatedWork W3192751986 @default.
- W2912272583 hasRelatedWork W3201345174 @default.
- W2912272583 hasRelatedWork W3203696624 @default.
- W2912272583 hasRelatedWork W3003520923 @default.
- W2912272583 hasRelatedWork W3082138134 @default.
- W2912272583 isParatext "false" @default.
- W2912272583 isRetracted "false" @default.
- W2912272583 magId "2912272583" @default.
- W2912272583 workType "article" @default.