FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3135377242> ?p ?o ?g. }

• W3135377242 abstract "In this paper we investigate how gradient-based algorithms such as gradient descent, (multi-pass) stochastic gradient descent, its persistent variant, and the Langevin algorithm navigate non-convex loss-landscapes and which of them is able to reach the best generalization error at limited sample complexity. We consider the loss landscape of the high-dimensional phase retrieval problem as a prototypical highly non-convex example. We observe that for phase retrieval the stochastic variants of gradient descent are able to reach perfect generalization for regions of control parameters where the gradient descent algorithm is not. We apply dynamical mean-field theory from statistical physics to characterize analytically the full trajectories of these algorithms in their continuous-time limit, with a warm start, and for large system sizes. We further unveil several intriguing properties of the landscape and the algorithms such as that the gradient descent can obtain better generalization properties from less informed initializations." @default.
• W3135377242 created "2021-03-15" @default.
• W3135377242 creator A5017614803 @default.
• W3135377242 creator A5054525547 @default.
• W3135377242 creator A5089268172 @default.
• W3135377242 date "2021-03-08" @default.
• W3135377242 modified "2023-09-27" @default.
• W3135377242 title "Stochasticity helps to navigate rough landscapes: comparing gradient-descent-based algorithms in the phase retrieval problem" @default.
• W3135377242 cites W1971565000 @default.
• W3135377242 cites W1977011475 @default.
• W3135377242 cites W1979344292 @default.
• W3135377242 cites W1988150410 @default.
• W3135377242 cites W2002963485 @default.
• W3135377242 cites W2026896061 @default.
• W3135377242 cites W2039263963 @default.
• W3135377242 cites W2044613770 @default.
• W3135377242 cites W2064512219 @default.
• W3135377242 cites W2074238984 @default.
• W3135377242 cites W2077647472 @default.
• W3135377242 cites W2167850383 @default.
• W3135377242 cites W2768267830 @default.
• W3135377242 cites W2804386825 @default.
• W3135377242 cites W2887611796 @default.
• W3135377242 cites W2899079412 @default.
• W3135377242 cites W2899900971 @default.
• W3135377242 cites W2918745211 @default.
• W3135377242 cites W2948488743 @default.
• W3135377242 cites W2962932339 @default.
• W3135377242 cites W2963519307 @default.
• W3135377242 cites W2963959597 @default.
• W3135377242 cites W2964072432 @default.
• W3135377242 cites W2995689697 @default.
• W3135377242 cites W2996603747 @default.
• W3135377242 cites W3000189159 @default.
• W3135377242 cites W3012806192 @default.
• W3135377242 cites W3023124831 @default.
• W3135377242 cites W3034840375 @default.
• W3135377242 cites W3034877436 @default.
• W3135377242 cites W3047952687 @default.
• W3135377242 cites W3101387215 @default.
• W3135377242 cites W3102766317 @default.
• W3135377242 cites W3104615236 @default.
• W3135377242 cites W3116143161 @default.
• W3135377242 cites W3120243577 @default.
• W3135377242 cites W3123272904 @default.
• W3135377242 hasPublicationYear "2021" @default.
• W3135377242 type Work @default.
• W3135377242 sameAs 3135377242 @default.
• W3135377242 citedByCount "0" @default.
• W3135377242 crossrefType "posted-content" @default.
• W3135377242 hasAuthorship W3135377242A5017614803 @default.
• W3135377242 hasAuthorship W3135377242A5054525547 @default.
• W3135377242 hasAuthorship W3135377242A5089268172 @default.
• W3135377242 hasConcept C105795698 @default.
• W3135377242 hasConcept C112680207 @default.
• W3135377242 hasConcept C11413529 @default.
• W3135377242 hasConcept C121332964 @default.
• W3135377242 hasConcept C126255220 @default.
• W3135377242 hasConcept C134306372 @default.
• W3135377242 hasConcept C151201525 @default.
• W3135377242 hasConcept C153258448 @default.
• W3135377242 hasConcept C153294291 @default.
• W3135377242 hasConcept C154945302 @default.
• W3135377242 hasConcept C177148314 @default.
• W3135377242 hasConcept C202444582 @default.
• W3135377242 hasConcept C206688291 @default.
• W3135377242 hasConcept C2524010 @default.
• W3135377242 hasConcept C2776637919 @default.
• W3135377242 hasConcept C2780004032 @default.
• W3135377242 hasConcept C28826006 @default.
• W3135377242 hasConcept C33923547 @default.
• W3135377242 hasConcept C41008148 @default.
• W3135377242 hasConcept C50644808 @default.
• W3135377242 hasConcept C9652623 @default.
• W3135377242 hasConceptScore W3135377242C105795698 @default.
• W3135377242 hasConceptScore W3135377242C112680207 @default.
• W3135377242 hasConceptScore W3135377242C11413529 @default.
• W3135377242 hasConceptScore W3135377242C121332964 @default.
• W3135377242 hasConceptScore W3135377242C126255220 @default.
• W3135377242 hasConceptScore W3135377242C134306372 @default.
• W3135377242 hasConceptScore W3135377242C151201525 @default.
• W3135377242 hasConceptScore W3135377242C153258448 @default.
• W3135377242 hasConceptScore W3135377242C153294291 @default.
• W3135377242 hasConceptScore W3135377242C154945302 @default.
• W3135377242 hasConceptScore W3135377242C177148314 @default.
• W3135377242 hasConceptScore W3135377242C202444582 @default.
• W3135377242 hasConceptScore W3135377242C206688291 @default.
• W3135377242 hasConceptScore W3135377242C2524010 @default.
• W3135377242 hasConceptScore W3135377242C2776637919 @default.
• W3135377242 hasConceptScore W3135377242C2780004032 @default.
• W3135377242 hasConceptScore W3135377242C28826006 @default.
• W3135377242 hasConceptScore W3135377242C33923547 @default.
• W3135377242 hasConceptScore W3135377242C41008148 @default.
• W3135377242 hasConceptScore W3135377242C50644808 @default.
• W3135377242 hasConceptScore W3135377242C9652623 @default.
• W3135377242 hasLocation W31353772421 @default.
• W3135377242 hasOpenAccess W3135377242 @default.
• W3135377242 hasPrimaryLocation W31353772421 @default.
• W3135377242 hasRelatedWork W2130314481 @default.
• W3135377242 hasRelatedWork W2166706236 @default.

• next