Matches in SemOpenAlex for { <https://semopenalex.org/work/W2617695214> ?p ?o ?g. }
- W2617695214 abstract "We develop a new theoretical framework to analyze the generalization error of deep learning, and derive a new fast learning rate for two representative algorithms: empirical risk minimization and Bayesian deep learning. The series of theoretical analyses of deep learning has revealed its high expressive power and universal approximation capability. Although these analyses are highly nonparametric, existing generalization error analyses have been developed mainly in a fixed dimensional parametric model. To compensate this gap, we develop an infinite dimensional model that is based on an integral form as performed in the analysis of the universal approximation capability. This allows us to define a reproducing kernel Hilbert space corresponding to each layer. Our point of view is to deal with the ordinary finite dimensional deep neural network as a finite approximation of the infinite dimensional one. The approximation error is evaluated by the degree of freedom of the reproducing kernel Hilbert space in each layer. To estimate a good finite dimensional model, we consider both of empirical risk minimization and Bayesian deep learning. We derive its generalization error bound and it is shown that there appears bias-variance trade-off in terms of the number of parameters of the finite dimensional approximation. We show that the optimal width of the internal layers can be determined through the degree of freedom and the convergence rate can be faster than $O(1/sqrt{n})$ rate which has been shown in the existing studies." @default.
- W2617695214 created "2017-06-05" @default.
- W2617695214 creator A5078812767 @default.
- W2617695214 date "2017-05-29" @default.
- W2617695214 modified "2023-09-25" @default.
- W2617695214 title "Fast learning rate of deep learning via a kernel perspective" @default.
- W2617695214 cites W1509689762 @default.
- W2617695214 cites W1543474109 @default.
- W2617695214 cites W1546851689 @default.
- W2617695214 cites W1608733719 @default.
- W2617695214 cites W1698155719 @default.
- W2617695214 cites W194417083 @default.
- W2617695214 cites W1986280275 @default.
- W2617695214 cites W1988115241 @default.
- W2617695214 cites W2007663459 @default.
- W2617695214 cites W2042799176 @default.
- W2617695214 cites W2059529070 @default.
- W2617695214 cites W2080872718 @default.
- W2617695214 cites W2095705004 @default.
- W2617695214 cites W2099579348 @default.
- W2617695214 cites W2100116519 @default.
- W2617695214 cites W2103496339 @default.
- W2617695214 cites W2115881141 @default.
- W2617695214 cites W2126417278 @default.
- W2617695214 cites W2167608136 @default.
- W2617695214 cites W2167784882 @default.
- W2617695214 cites W2423689290 @default.
- W2617695214 cites W2462875204 @default.
- W2617695214 cites W2557086248 @default.
- W2617695214 cites W2949336664 @default.
- W2617695214 cites W2949578333 @default.
- W2617695214 cites W2950177356 @default.
- W2617695214 cites W2951266961 @default.
- W2617695214 cites W2951603627 @default.
- W2617695214 cites W2951654389 @default.
- W2617695214 cites W2951886858 @default.
- W2617695214 cites W2962742960 @default.
- W2617695214 cites W2962845550 @default.
- W2617695214 cites W2962857907 @default.
- W2617695214 cites W2963534251 @default.
- W2617695214 cites W2963709899 @default.
- W2617695214 cites W2963766931 @default.
- W2617695214 cites W2964059111 @default.
- W2617695214 cites W3105849782 @default.
- W2617695214 cites W600157613 @default.
- W2617695214 doi "https://doi.org/10.48550/arxiv.1705.10182" @default.
- W2617695214 hasPublicationYear "2017" @default.
- W2617695214 type Work @default.
- W2617695214 sameAs 2617695214 @default.
- W2617695214 citedByCount "4" @default.
- W2617695214 countsByYear W26176952142018 @default.
- W2617695214 countsByYear W26176952142019 @default.
- W2617695214 crossrefType "posted-content" @default.
- W2617695214 hasAuthorship W2617695214A5078812767 @default.
- W2617695214 hasBestOaLocation W26176952141 @default.
- W2617695214 hasConcept C107321475 @default.
- W2617695214 hasConcept C108583219 @default.
- W2617695214 hasConcept C11413529 @default.
- W2617695214 hasConcept C118615104 @default.
- W2617695214 hasConcept C122383733 @default.
- W2617695214 hasConcept C126255220 @default.
- W2617695214 hasConcept C127162648 @default.
- W2617695214 hasConcept C134306372 @default.
- W2617695214 hasConcept C154945302 @default.
- W2617695214 hasConcept C177148314 @default.
- W2617695214 hasConcept C28826006 @default.
- W2617695214 hasConcept C31258907 @default.
- W2617695214 hasConcept C33923547 @default.
- W2617695214 hasConcept C41008148 @default.
- W2617695214 hasConcept C50644808 @default.
- W2617695214 hasConcept C57869625 @default.
- W2617695214 hasConcept C62799726 @default.
- W2617695214 hasConcept C74193536 @default.
- W2617695214 hasConcept C80884492 @default.
- W2617695214 hasConceptScore W2617695214C107321475 @default.
- W2617695214 hasConceptScore W2617695214C108583219 @default.
- W2617695214 hasConceptScore W2617695214C11413529 @default.
- W2617695214 hasConceptScore W2617695214C118615104 @default.
- W2617695214 hasConceptScore W2617695214C122383733 @default.
- W2617695214 hasConceptScore W2617695214C126255220 @default.
- W2617695214 hasConceptScore W2617695214C127162648 @default.
- W2617695214 hasConceptScore W2617695214C134306372 @default.
- W2617695214 hasConceptScore W2617695214C154945302 @default.
- W2617695214 hasConceptScore W2617695214C177148314 @default.
- W2617695214 hasConceptScore W2617695214C28826006 @default.
- W2617695214 hasConceptScore W2617695214C31258907 @default.
- W2617695214 hasConceptScore W2617695214C33923547 @default.
- W2617695214 hasConceptScore W2617695214C41008148 @default.
- W2617695214 hasConceptScore W2617695214C50644808 @default.
- W2617695214 hasConceptScore W2617695214C57869625 @default.
- W2617695214 hasConceptScore W2617695214C62799726 @default.
- W2617695214 hasConceptScore W2617695214C74193536 @default.
- W2617695214 hasConceptScore W2617695214C80884492 @default.
- W2617695214 hasLocation W26176952141 @default.
- W2617695214 hasOpenAccess W2617695214 @default.
- W2617695214 hasPrimaryLocation W26176952141 @default.
- W2617695214 hasRelatedWork W183296071 @default.
- W2617695214 hasRelatedWork W2058007550 @default.
- W2617695214 hasRelatedWork W2087986550 @default.
- W2617695214 hasRelatedWork W2341769498 @default.