Matches in SemOpenAlex for { <https://semopenalex.org/work/W2970332347> ?p ?o ?g. }
Showing items 1 to 84 of
84
with 100 items per page.
- W2970332347 endingPage "10845" @default.
- W2970332347 startingPage "10835" @default.
- W2970332347 abstract "We study the training and generalization of deep neural networks (DNNs) in the over-parameterized regime, where the network width (i.e., number of hidden nodes per layer) is much larger than the number of training data points. We show that, the expected $0$-$1$ loss of a wide enough ReLU network trained with stochastic gradient descent (SGD) and random initialization can be bounded by the training loss of a random feature model induced by the network gradient at initialization, which we call a textit{neural tangent random feature} (NTRF) model. For data distributions that can be classified by NTRF model with sufficiently small error, our result yields a generalization error bound in the order of $tilde{mathcal{O}}(n^{-1/2})$ that is independent of the network width. Our result is more general and sharper than many existing generalization error bounds for over-parameterized neural networks. In addition, we establish a strong connection between our generalization error bound and the neural tangent kernel (NTK) proposed in recent work." @default.
- W2970332347 created "2019-09-05" @default.
- W2970332347 creator A5051448391 @default.
- W2970332347 creator A5086329948 @default.
- W2970332347 date "2019-01-01" @default.
- W2970332347 modified "2023-10-18" @default.
- W2970332347 title "Generalization Bounds of Stochastic Gradient Descent for Wide and Deep Neural Networks" @default.
- W2970332347 hasPublicationYear "2019" @default.
- W2970332347 type Work @default.
- W2970332347 sameAs 2970332347 @default.
- W2970332347 citedByCount "95" @default.
- W2970332347 countsByYear W29703323472019 @default.
- W2970332347 countsByYear W29703323472020 @default.
- W2970332347 countsByYear W29703323472021 @default.
- W2970332347 countsByYear W29703323472022 @default.
- W2970332347 crossrefType "proceedings-article" @default.
- W2970332347 hasAuthorship W2970332347A5051448391 @default.
- W2970332347 hasAuthorship W2970332347A5086329948 @default.
- W2970332347 hasConcept C11413529 @default.
- W2970332347 hasConcept C114466953 @default.
- W2970332347 hasConcept C134306372 @default.
- W2970332347 hasConcept C138187205 @default.
- W2970332347 hasConcept C147168706 @default.
- W2970332347 hasConcept C153258448 @default.
- W2970332347 hasConcept C154945302 @default.
- W2970332347 hasConcept C165464430 @default.
- W2970332347 hasConcept C177148314 @default.
- W2970332347 hasConcept C199360897 @default.
- W2970332347 hasConcept C206688291 @default.
- W2970332347 hasConcept C2524010 @default.
- W2970332347 hasConcept C33923547 @default.
- W2970332347 hasConcept C34388435 @default.
- W2970332347 hasConcept C41008148 @default.
- W2970332347 hasConcept C50644808 @default.
- W2970332347 hasConcept C77553402 @default.
- W2970332347 hasConcept C86582703 @default.
- W2970332347 hasConceptScore W2970332347C11413529 @default.
- W2970332347 hasConceptScore W2970332347C114466953 @default.
- W2970332347 hasConceptScore W2970332347C134306372 @default.
- W2970332347 hasConceptScore W2970332347C138187205 @default.
- W2970332347 hasConceptScore W2970332347C147168706 @default.
- W2970332347 hasConceptScore W2970332347C153258448 @default.
- W2970332347 hasConceptScore W2970332347C154945302 @default.
- W2970332347 hasConceptScore W2970332347C165464430 @default.
- W2970332347 hasConceptScore W2970332347C177148314 @default.
- W2970332347 hasConceptScore W2970332347C199360897 @default.
- W2970332347 hasConceptScore W2970332347C206688291 @default.
- W2970332347 hasConceptScore W2970332347C2524010 @default.
- W2970332347 hasConceptScore W2970332347C33923547 @default.
- W2970332347 hasConceptScore W2970332347C34388435 @default.
- W2970332347 hasConceptScore W2970332347C41008148 @default.
- W2970332347 hasConceptScore W2970332347C50644808 @default.
- W2970332347 hasConceptScore W2970332347C77553402 @default.
- W2970332347 hasConceptScore W2970332347C86582703 @default.
- W2970332347 hasLocation W29703323471 @default.
- W2970332347 hasOpenAccess W2970332347 @default.
- W2970332347 hasPrimaryLocation W29703323471 @default.
- W2970332347 hasRelatedWork W2194775991 @default.
- W2970332347 hasRelatedWork W2809090039 @default.
- W2970332347 hasRelatedWork W2886067286 @default.
- W2970332347 hasRelatedWork W2899748887 @default.
- W2970332347 hasRelatedWork W2900959181 @default.
- W2970332347 hasRelatedWork W2913473169 @default.
- W2970332347 hasRelatedWork W2946840143 @default.
- W2970332347 hasRelatedWork W2952204734 @default.
- W2970332347 hasRelatedWork W2962698540 @default.
- W2970332347 hasRelatedWork W2963239103 @default.
- W2970332347 hasRelatedWork W2964161337 @default.
- W2970332347 hasRelatedWork W2970217468 @default.
- W2970332347 hasRelatedWork W2970239942 @default.
- W2970332347 hasRelatedWork W2970330753 @default.
- W2970332347 hasRelatedWork W2970443625 @default.
- W2970332347 hasRelatedWork W2971043187 @default.
- W2970332347 hasRelatedWork W2971169274 @default.
- W2970332347 hasRelatedWork W2996168800 @default.
- W2970332347 hasRelatedWork W3137695714 @default.
- W2970332347 hasRelatedWork W3152114226 @default.
- W2970332347 hasVolume "32" @default.
- W2970332347 isParatext "false" @default.
- W2970332347 isRetracted "false" @default.
- W2970332347 magId "2970332347" @default.
- W2970332347 workType "article" @default.