Matches in SemOpenAlex for { <https://semopenalex.org/work/W2768041186> ?p ?o ?g. }
Showing items 1 to 95 of
95
with 100 items per page.
- W2768041186 abstract "In this paper, we consider parameter recovery for non-overlapping convolutional neural networks (CNNs) with multiple kernels. We show that when the inputs follow Gaussian distribution and the sample size is sufficiently large, the squared loss of such CNNs is $mathit{~locally~strongly~convex}$ in a basin of attraction near the global optima for most popular activation functions, like ReLU, Leaky ReLU, Squared ReLU, Sigmoid and Tanh. The required sample complexity is proportional to the dimension of the input and polynomial in the number of kernels and a condition number of the parameters. We also show that tensor methods are able to initialize the parameters to the local strong convex region. Hence, for most smooth activations, gradient descent following tensor initialization is guaranteed to converge to the global optimal with time that is linear in input dimension, logarithmic in precision and polynomial in other factors. To the best of our knowledge, this is the first work that provides recovery guarantees for CNNs with multiple kernels under polynomial sample and computational complexities." @default.
- W2768041186 created "2017-11-17" @default.
- W2768041186 creator A5017721496 @default.
- W2768041186 creator A5063459703 @default.
- W2768041186 creator A5091070408 @default.
- W2768041186 date "2017-11-08" @default.
- W2768041186 modified "2023-09-27" @default.
- W2768041186 title "Learning Non-overlapping Convolutional Neural Networks with Multiple Kernels" @default.
- W2768041186 cites W2433379750 @default.
- W2768041186 cites W2553316781 @default.
- W2768041186 cites W2609713339 @default.
- W2768041186 cites W2953281724 @default.
- W2768041186 hasPublicationYear "2017" @default.
- W2768041186 type Work @default.
- W2768041186 sameAs 2768041186 @default.
- W2768041186 citedByCount "33" @default.
- W2768041186 countsByYear W27680411862018 @default.
- W2768041186 countsByYear W27680411862019 @default.
- W2768041186 countsByYear W27680411862020 @default.
- W2768041186 countsByYear W27680411862021 @default.
- W2768041186 crossrefType "posted-content" @default.
- W2768041186 hasAuthorship W2768041186A5017721496 @default.
- W2768041186 hasAuthorship W2768041186A5063459703 @default.
- W2768041186 hasAuthorship W2768041186A5091070408 @default.
- W2768041186 hasConcept C11413529 @default.
- W2768041186 hasConcept C114466953 @default.
- W2768041186 hasConcept C114614502 @default.
- W2768041186 hasConcept C121332964 @default.
- W2768041186 hasConcept C126255220 @default.
- W2768041186 hasConcept C134306372 @default.
- W2768041186 hasConcept C153258448 @default.
- W2768041186 hasConcept C154945302 @default.
- W2768041186 hasConcept C155281189 @default.
- W2768041186 hasConcept C163716315 @default.
- W2768041186 hasConcept C199360897 @default.
- W2768041186 hasConcept C202444582 @default.
- W2768041186 hasConcept C28826006 @default.
- W2768041186 hasConcept C33676613 @default.
- W2768041186 hasConcept C33923547 @default.
- W2768041186 hasConcept C39927690 @default.
- W2768041186 hasConcept C41008148 @default.
- W2768041186 hasConcept C50644808 @default.
- W2768041186 hasConcept C62520636 @default.
- W2768041186 hasConcept C81363708 @default.
- W2768041186 hasConcept C81388566 @default.
- W2768041186 hasConcept C90119067 @default.
- W2768041186 hasConceptScore W2768041186C11413529 @default.
- W2768041186 hasConceptScore W2768041186C114466953 @default.
- W2768041186 hasConceptScore W2768041186C114614502 @default.
- W2768041186 hasConceptScore W2768041186C121332964 @default.
- W2768041186 hasConceptScore W2768041186C126255220 @default.
- W2768041186 hasConceptScore W2768041186C134306372 @default.
- W2768041186 hasConceptScore W2768041186C153258448 @default.
- W2768041186 hasConceptScore W2768041186C154945302 @default.
- W2768041186 hasConceptScore W2768041186C155281189 @default.
- W2768041186 hasConceptScore W2768041186C163716315 @default.
- W2768041186 hasConceptScore W2768041186C199360897 @default.
- W2768041186 hasConceptScore W2768041186C202444582 @default.
- W2768041186 hasConceptScore W2768041186C28826006 @default.
- W2768041186 hasConceptScore W2768041186C33676613 @default.
- W2768041186 hasConceptScore W2768041186C33923547 @default.
- W2768041186 hasConceptScore W2768041186C39927690 @default.
- W2768041186 hasConceptScore W2768041186C41008148 @default.
- W2768041186 hasConceptScore W2768041186C50644808 @default.
- W2768041186 hasConceptScore W2768041186C62520636 @default.
- W2768041186 hasConceptScore W2768041186C81363708 @default.
- W2768041186 hasConceptScore W2768041186C81388566 @default.
- W2768041186 hasConceptScore W2768041186C90119067 @default.
- W2768041186 hasLocation W27680411861 @default.
- W2768041186 hasOpenAccess W2768041186 @default.
- W2768041186 hasPrimaryLocation W27680411861 @default.
- W2768041186 hasRelatedWork W1839868949 @default.
- W2768041186 hasRelatedWork W2194775991 @default.
- W2768041186 hasRelatedWork W2257979135 @default.
- W2768041186 hasRelatedWork W2591714514 @default.
- W2768041186 hasRelatedWork W2758053331 @default.
- W2768041186 hasRelatedWork W2809090039 @default.
- W2768041186 hasRelatedWork W2886067286 @default.
- W2768041186 hasRelatedWork W2894604724 @default.
- W2768041186 hasRelatedWork W2899748887 @default.
- W2768041186 hasRelatedWork W2962698540 @default.
- W2768041186 hasRelatedWork W2962767131 @default.
- W2768041186 hasRelatedWork W2962930448 @default.
- W2768041186 hasRelatedWork W2963211922 @default.
- W2768041186 hasRelatedWork W2963383839 @default.
- W2768041186 hasRelatedWork W2963519230 @default.
- W2768041186 hasRelatedWork W2963623651 @default.
- W2768041186 hasRelatedWork W2963744427 @default.
- W2768041186 hasRelatedWork W2963827833 @default.
- W2768041186 hasRelatedWork W2965497096 @default.
- W2768041186 hasRelatedWork W2991290085 @default.
- W2768041186 isParatext "false" @default.
- W2768041186 isRetracted "false" @default.
- W2768041186 magId "2768041186" @default.
- W2768041186 workType "article" @default.