Matches in SemOpenAlex for { <https://semopenalex.org/work/W2912004605> ?p ?o ?g. }
- W2912004605 abstract "Loss functions with a large number of saddle points are one of the main obstacles to training many modern machine learning models. Gradient descent (GD) is a fundamental algorithm for machine learning and converges to a saddle point for certain initial data. We call the region formed by these initial values the For quadratic functions, GD converges to a saddle point if the initial data is in a subspace of up to n-1 dimensions. In this paper, we prove that a small modification of the recently proposed Laplacian smoothing gradient descent (LSGD) [Osher, et al., arXiv:1806.06317] contributes to avoiding saddle points without sacrificing the convergence rate of GD. In particular, we show that the dimension of the LSGD's attraction region is at most floor((n-1)/2) for a class of quadratic functions which is significantly smaller than GD's (n-1)-dimensional attraction region." @default.
- W2912004605 created "2019-02-21" @default.
- W2912004605 creator A5002037883 @default.
- W2912004605 creator A5026542884 @default.
- W2912004605 creator A5036026030 @default.
- W2912004605 date "2019-01-21" @default.
- W2912004605 modified "2023-09-27" @default.
- W2912004605 title "A Deterministic Approach to Avoid Saddle Points." @default.
- W2912004605 cites W2146989110 @default.
- W2912004605 cites W2283214199 @default.
- W2912004605 cites W2396019715 @default.
- W2912004605 cites W2398500126 @default.
- W2912004605 cites W2561193401 @default.
- W2912004605 cites W2597410408 @default.
- W2912004605 cites W2609037894 @default.
- W2912004605 cites W2963416883 @default.
- W2912004605 cites W2963446085 @default.
- W2912004605 cites W2963487351 @default.
- W2912004605 cites W2964106499 @default.
- W2912004605 hasPublicationYear "2019" @default.
- W2912004605 type Work @default.
- W2912004605 sameAs 2912004605 @default.
- W2912004605 citedByCount "2" @default.
- W2912004605 countsByYear W29120046052018 @default.
- W2912004605 countsByYear W29120046052019 @default.
- W2912004605 crossrefType "posted-content" @default.
- W2912004605 hasAuthorship W2912004605A5002037883 @default.
- W2912004605 hasAuthorship W2912004605A5026542884 @default.
- W2912004605 hasAuthorship W2912004605A5036026030 @default.
- W2912004605 hasConcept C105795698 @default.
- W2912004605 hasConcept C11413529 @default.
- W2912004605 hasConcept C126255220 @default.
- W2912004605 hasConcept C127162648 @default.
- W2912004605 hasConcept C129844170 @default.
- W2912004605 hasConcept C134306372 @default.
- W2912004605 hasConcept C153258448 @default.
- W2912004605 hasConcept C154945302 @default.
- W2912004605 hasConcept C162324750 @default.
- W2912004605 hasConcept C165700671 @default.
- W2912004605 hasConcept C202444582 @default.
- W2912004605 hasConcept C21080849 @default.
- W2912004605 hasConcept C2524010 @default.
- W2912004605 hasConcept C2681867 @default.
- W2912004605 hasConcept C2777127463 @default.
- W2912004605 hasConcept C2777303404 @default.
- W2912004605 hasConcept C28826006 @default.
- W2912004605 hasConcept C31258907 @default.
- W2912004605 hasConcept C32834561 @default.
- W2912004605 hasConcept C33676613 @default.
- W2912004605 hasConcept C33923547 @default.
- W2912004605 hasConcept C3770464 @default.
- W2912004605 hasConcept C41008148 @default.
- W2912004605 hasConcept C50522688 @default.
- W2912004605 hasConcept C50644808 @default.
- W2912004605 hasConcept C57869625 @default.
- W2912004605 hasConceptScore W2912004605C105795698 @default.
- W2912004605 hasConceptScore W2912004605C11413529 @default.
- W2912004605 hasConceptScore W2912004605C126255220 @default.
- W2912004605 hasConceptScore W2912004605C127162648 @default.
- W2912004605 hasConceptScore W2912004605C129844170 @default.
- W2912004605 hasConceptScore W2912004605C134306372 @default.
- W2912004605 hasConceptScore W2912004605C153258448 @default.
- W2912004605 hasConceptScore W2912004605C154945302 @default.
- W2912004605 hasConceptScore W2912004605C162324750 @default.
- W2912004605 hasConceptScore W2912004605C165700671 @default.
- W2912004605 hasConceptScore W2912004605C202444582 @default.
- W2912004605 hasConceptScore W2912004605C21080849 @default.
- W2912004605 hasConceptScore W2912004605C2524010 @default.
- W2912004605 hasConceptScore W2912004605C2681867 @default.
- W2912004605 hasConceptScore W2912004605C2777127463 @default.
- W2912004605 hasConceptScore W2912004605C2777303404 @default.
- W2912004605 hasConceptScore W2912004605C28826006 @default.
- W2912004605 hasConceptScore W2912004605C31258907 @default.
- W2912004605 hasConceptScore W2912004605C32834561 @default.
- W2912004605 hasConceptScore W2912004605C33676613 @default.
- W2912004605 hasConceptScore W2912004605C33923547 @default.
- W2912004605 hasConceptScore W2912004605C3770464 @default.
- W2912004605 hasConceptScore W2912004605C41008148 @default.
- W2912004605 hasConceptScore W2912004605C50522688 @default.
- W2912004605 hasConceptScore W2912004605C50644808 @default.
- W2912004605 hasConceptScore W2912004605C57869625 @default.
- W2912004605 hasLocation W29120046051 @default.
- W2912004605 hasOpenAccess W2912004605 @default.
- W2912004605 hasPrimaryLocation W29120046051 @default.
- W2912004605 hasRelatedWork W1592396907 @default.
- W2912004605 hasRelatedWork W1657275981 @default.
- W2912004605 hasRelatedWork W2078853581 @default.
- W2912004605 hasRelatedWork W2159503542 @default.
- W2912004605 hasRelatedWork W2889795689 @default.
- W2912004605 hasRelatedWork W2892569780 @default.
- W2912004605 hasRelatedWork W2908088598 @default.
- W2912004605 hasRelatedWork W2949746748 @default.
- W2912004605 hasRelatedWork W2965341779 @default.
- W2912004605 hasRelatedWork W2971250964 @default.
- W2912004605 hasRelatedWork W3006924802 @default.
- W2912004605 hasRelatedWork W3025065608 @default.
- W2912004605 hasRelatedWork W3105204313 @default.
- W2912004605 hasRelatedWork W3133514242 @default.
- W2912004605 hasRelatedWork W3136044007 @default.
- W2912004605 hasRelatedWork W3147777701 @default.