Matches in SemOpenAlex for { <https://semopenalex.org/work/W3043999652> ?p ?o ?g. }
Showing items 1 to 87 of
87
with 100 items per page.
- W3043999652 abstract "We study the statistical properties of the dynamic trajectory of stochastic gradient descent (SGD). We approximate the mini-batch SGD and the momentum SGD as stochastic differential equations. We exploit the continuous formulation of SDE and the theory of Fokker–Planck equations to develop new results on the escaping phenomenon and the relationship with large batch and sharp minima. In particular, we find that the stochastic process solution tends to converge to flatter minima regardless of the batch size in the asymptotic regime. However, the convergence rate is rigorously proven to depend on the batch size. These results are validated empirically with various datasets and models." @default.
- W3043999652 created "2020-07-29" @default.
- W3043999652 creator A5000889358 @default.
- W3043999652 creator A5063403085 @default.
- W3043999652 date "2020-07-26" @default.
- W3043999652 modified "2023-10-17" @default.
- W3043999652 title "On Large Batch Training and Sharp Minima: A Fokker–Planck Perspective" @default.
- W3043999652 cites W1980287119 @default.
- W3043999652 cites W2040060987 @default.
- W3043999652 cites W2097209344 @default.
- W3043999652 cites W2194775991 @default.
- W3043999652 cites W2798619491 @default.
- W3043999652 cites W2963177640 @default.
- W3043999652 cites W2963433607 @default.
- W3043999652 cites W3099529713 @default.
- W3043999652 cites W4238209390 @default.
- W3043999652 cites W4253907471 @default.
- W3043999652 doi "https://doi.org/10.1007/s42519-020-00120-9" @default.
- W3043999652 hasPublicationYear "2020" @default.
- W3043999652 type Work @default.
- W3043999652 sameAs 3043999652 @default.
- W3043999652 citedByCount "3" @default.
- W3043999652 countsByYear W30439996522021 @default.
- W3043999652 crossrefType "journal-article" @default.
- W3043999652 hasAuthorship W3043999652A5000889358 @default.
- W3043999652 hasAuthorship W3043999652A5063403085 @default.
- W3043999652 hasBestOaLocation W30439996522 @default.
- W3043999652 hasConcept C10138342 @default.
- W3043999652 hasConcept C121332964 @default.
- W3043999652 hasConcept C121864883 @default.
- W3043999652 hasConcept C126255220 @default.
- W3043999652 hasConcept C1276947 @default.
- W3043999652 hasConcept C134306372 @default.
- W3043999652 hasConcept C13662910 @default.
- W3043999652 hasConcept C154945302 @default.
- W3043999652 hasConcept C162324750 @default.
- W3043999652 hasConcept C186633575 @default.
- W3043999652 hasConcept C206688291 @default.
- W3043999652 hasConcept C2777303404 @default.
- W3043999652 hasConcept C28826006 @default.
- W3043999652 hasConcept C33923547 @default.
- W3043999652 hasConcept C41008148 @default.
- W3043999652 hasConcept C50522688 @default.
- W3043999652 hasConcept C50644808 @default.
- W3043999652 hasConcept C51955184 @default.
- W3043999652 hasConcept C60718061 @default.
- W3043999652 hasConcept C69123182 @default.
- W3043999652 hasConcept C78045399 @default.
- W3043999652 hasConceptScore W3043999652C10138342 @default.
- W3043999652 hasConceptScore W3043999652C121332964 @default.
- W3043999652 hasConceptScore W3043999652C121864883 @default.
- W3043999652 hasConceptScore W3043999652C126255220 @default.
- W3043999652 hasConceptScore W3043999652C1276947 @default.
- W3043999652 hasConceptScore W3043999652C134306372 @default.
- W3043999652 hasConceptScore W3043999652C13662910 @default.
- W3043999652 hasConceptScore W3043999652C154945302 @default.
- W3043999652 hasConceptScore W3043999652C162324750 @default.
- W3043999652 hasConceptScore W3043999652C186633575 @default.
- W3043999652 hasConceptScore W3043999652C206688291 @default.
- W3043999652 hasConceptScore W3043999652C2777303404 @default.
- W3043999652 hasConceptScore W3043999652C28826006 @default.
- W3043999652 hasConceptScore W3043999652C33923547 @default.
- W3043999652 hasConceptScore W3043999652C41008148 @default.
- W3043999652 hasConceptScore W3043999652C50522688 @default.
- W3043999652 hasConceptScore W3043999652C50644808 @default.
- W3043999652 hasConceptScore W3043999652C51955184 @default.
- W3043999652 hasConceptScore W3043999652C60718061 @default.
- W3043999652 hasConceptScore W3043999652C69123182 @default.
- W3043999652 hasConceptScore W3043999652C78045399 @default.
- W3043999652 hasLocation W30439996521 @default.
- W3043999652 hasLocation W30439996522 @default.
- W3043999652 hasOpenAccess W3043999652 @default.
- W3043999652 hasPrimaryLocation W30439996521 @default.
- W3043999652 hasRelatedWork W17442800 @default.
- W3043999652 hasRelatedWork W20428684 @default.
- W3043999652 hasRelatedWork W20449131 @default.
- W3043999652 hasRelatedWork W28235950 @default.
- W3043999652 hasRelatedWork W29127251 @default.
- W3043999652 hasRelatedWork W32780757 @default.
- W3043999652 hasRelatedWork W33732252 @default.
- W3043999652 hasRelatedWork W34224590 @default.
- W3043999652 hasRelatedWork W39130618 @default.
- W3043999652 hasRelatedWork W39667797 @default.
- W3043999652 isParatext "false" @default.
- W3043999652 isRetracted "false" @default.
- W3043999652 magId "3043999652" @default.
- W3043999652 workType "article" @default.