FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2899790086> ?p ?o ?g. }

• W2899790086 abstract "Gradient descent finds a global minimum in training deep neural networks despite the objective function being non-convex. The current paper proves gradient descent achieves zero training loss in polynomial time for a deep over-parameterized neural network with residual connections (ResNet). Our analysis relies on the particular structure of the Gram matrix induced by the neural network architecture. This structure allows us to show the Gram matrix is stable throughout the training process and this stability implies the global optimality of the gradient descent algorithm. We further extend our analysis to deep residual convolutional neural networks and obtain a similar convergence result." @default.
• W2899790086 created "2018-11-16" @default.
• W2899790086 creator A5024083506 @default.
• W2899790086 creator A5033061754 @default.
• W2899790086 creator A5055723755 @default.
• W2899790086 creator A5059740024 @default.
• W2899790086 creator A5068907670 @default.
• W2899790086 date "2018-11-09" @default.
• W2899790086 modified "2023-09-23" @default.
• W2899790086 title "Gradient Descent Finds Global Minima of Deep Neural Networks" @default.
• W2899790086 cites W1508863780 @default.
• W2899790086 cites W2113517874 @default.
• W2899790086 cites W2194775991 @default.
• W2899790086 cites W2399994860 @default.
• W2899790086 cites W2474090883 @default.
• W2899790086 cites W2591714514 @default.
• W2899790086 cites W2593958421 @default.
• W2899790086 cites W2625063094 @default.
• W2899790086 cites W2758053331 @default.
• W2899790086 cites W2768041186 @default.
• W2899790086 cites W2785626633 @default.
• W2899790086 cites W2788997738 @default.
• W2899790086 cites W2795605442 @default.
• W2899790086 cites W2798826368 @default.
• W2899790086 cites W2798986185 @default.
• W2899790086 cites W2809090039 @default.
• W2899790086 cites W2886067286 @default.
• W2899790086 cites W2894604724 @default.
• W2899790086 cites W2897080984 @default.
• W2899790086 cites W2899748887 @default.
• W2899790086 cites W2900959181 @default.
• W2899790086 cites W2904838594 @default.
• W2899790086 cites W2911867426 @default.
• W2899790086 cites W2951934643 @default.
• W2899790086 cites W2952574409 @default.
• W2899790086 cites W2962930448 @default.
• W2899790086 cites W2963092340 @default.
• W2899790086 cites W2963095610 @default.
• W2899790086 cites W2963097630 @default.
• W2899790086 cites W2963326517 @default.
• W2899790086 cites W2963383839 @default.
• W2899790086 cites W2963416883 @default.
• W2899790086 cites W2963417959 @default.
• W2899790086 cites W2963427613 @default.
• W2899790086 cites W2963446085 @default.
• W2899790086 cites W2963519230 @default.
• W2899790086 cites W2963623651 @default.
• W2899790086 cites W2963651774 @default.
• W2899790086 cites W2963744427 @default.
• W2899790086 cites W2963827833 @default.
• W2899790086 cites W2963982496 @default.
• W2899790086 cites W2964052793 @default.
• W2899790086 cites W2964106499 @default.
• W2899790086 cites W2970330753 @default.
• W2899790086 cites W2991290085 @default.
• W2899790086 cites W3137695714 @default.
• W2899790086 cites W813605148 @default.
• W2899790086 hasPublicationYear "2018" @default.
• W2899790086 type Work @default.
• W2899790086 sameAs 2899790086 @default.
• W2899790086 citedByCount "21" @default.
• W2899790086 countsByYear W28997900862019 @default.
• W2899790086 countsByYear W28997900862020 @default.
• W2899790086 countsByYear W28997900862021 @default.
• W2899790086 crossrefType "posted-content" @default.
• W2899790086 hasAuthorship W2899790086A5024083506 @default.
• W2899790086 hasAuthorship W2899790086A5033061754 @default.
• W2899790086 hasAuthorship W2899790086A5055723755 @default.
• W2899790086 hasAuthorship W2899790086A5059740024 @default.
• W2899790086 hasAuthorship W2899790086A5068907670 @default.
• W2899790086 hasConcept C108583219 @default.
• W2899790086 hasConcept C11413529 @default.
• W2899790086 hasConcept C134306372 @default.
• W2899790086 hasConcept C153258448 @default.
• W2899790086 hasConcept C154945302 @default.
• W2899790086 hasConcept C155512373 @default.
• W2899790086 hasConcept C162324750 @default.
• W2899790086 hasConcept C165464430 @default.
• W2899790086 hasConcept C186633575 @default.
• W2899790086 hasConcept C2777303404 @default.
• W2899790086 hasConcept C33923547 @default.
• W2899790086 hasConcept C41008148 @default.
• W2899790086 hasConcept C50522688 @default.
• W2899790086 hasConcept C50644808 @default.
• W2899790086 hasConcept C81363708 @default.
• W2899790086 hasConceptScore W2899790086C108583219 @default.
• W2899790086 hasConceptScore W2899790086C11413529 @default.
• W2899790086 hasConceptScore W2899790086C134306372 @default.
• W2899790086 hasConceptScore W2899790086C153258448 @default.
• W2899790086 hasConceptScore W2899790086C154945302 @default.
• W2899790086 hasConceptScore W2899790086C155512373 @default.
• W2899790086 hasConceptScore W2899790086C162324750 @default.
• W2899790086 hasConceptScore W2899790086C165464430 @default.
• W2899790086 hasConceptScore W2899790086C186633575 @default.
• W2899790086 hasConceptScore W2899790086C2777303404 @default.
• W2899790086 hasConceptScore W2899790086C33923547 @default.
• W2899790086 hasConceptScore W2899790086C41008148 @default.
• W2899790086 hasConceptScore W2899790086C50522688 @default.
• W2899790086 hasConceptScore W2899790086C50644808 @default.
• W2899790086 hasConceptScore W2899790086C81363708 @default.

• next