FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3005703306> ?p ?o ?g. }

• W3005703306 abstract "In supervised learning using kernel methods, we encounter a large-scale finite-sum minimization over a reproducing kernel Hilbert space (RKHS). Often times large-scale finite-sum problems can be solved using efficient variants of Newton's method where the Hessian is approximated via sub-samples. In RKHS, however, the dependence of the penalty function to kernel makes standard sub-sampling approaches inapplicable, since the gram matrix is not readily available in a low-rank form. In this paper, we observe that for this class of problems, one can naturally use kernel approximation to speed up the Newton's method. Focusing on randomized features for kernel approximation, we provide a novel second-order algorithm that enjoys local superlinear convergence and global convergence in the high probability sense. The key to our analysis is showing that the approximated Hessian via random features preserves the spectrum of the original Hessian. We provide numerical experiments verifying the efficiency of our approach, compared to variants of sub-sampling methods." @default.
• W3005703306 created "2020-02-24" @default.
• W3005703306 creator A5040922600 @default.
• W3005703306 creator A5087313193 @default.
• W3005703306 date "2020-02-12" @default.
• W3005703306 modified "2023-09-27" @default.
• W3005703306 title "A Random-Feature Based Newton Method for Empirical Risk Minimization in Reproducing Kernel Hilbert Space." @default.
• W3005703306 cites W1491468723 @default.
• W3005703306 cites W1491622225 @default.
• W3005703306 cites W1510073064 @default.
• W3005703306 cites W1592294486 @default.
• W3005703306 cites W1622263187 @default.
• W3005703306 cites W196761320 @default.
• W3005703306 cites W1988813039 @default.
• W3005703306 cites W1991083751 @default.
• W3005703306 cites W2005136695 @default.
• W3005703306 cites W2015627422 @default.
• W3005703306 cites W2038210983 @default.
• W3005703306 cites W2042173174 @default.
• W3005703306 cites W2056641273 @default.
• W3005703306 cites W2061570747 @default.
• W3005703306 cites W2078409719 @default.
• W3005703306 cites W2103346443 @default.
• W3005703306 cites W2118297240 @default.
• W3005703306 cites W2123395972 @default.
• W3005703306 cites W2137515395 @default.
• W3005703306 cites W2144902422 @default.
• W3005703306 cites W2467074172 @default.
• W3005703306 cites W2583710113 @default.
• W3005703306 cites W2616038372 @default.
• W3005703306 cites W2727020975 @default.
• W3005703306 cites W2804140211 @default.
• W3005703306 cites W2900789157 @default.
• W3005703306 cites W2962840796 @default.
• W3005703306 cites W2962961534 @default.
• W3005703306 cites W2963060476 @default.
• W3005703306 cites W2963397933 @default.
• W3005703306 cites W2963433607 @default.
• W3005703306 cites W2963459001 @default.
• W3005703306 cites W2963483271 @default.
• W3005703306 cites W2963755879 @default.
• W3005703306 cites W2963807877 @default.
• W3005703306 cites W2963813563 @default.
• W3005703306 cites W2964063026 @default.
• W3005703306 cites W2964089577 @default.
• W3005703306 hasPublicationYear "2020" @default.
• W3005703306 type Work @default.
• W3005703306 sameAs 3005703306 @default.
• W3005703306 citedByCount "0" @default.
• W3005703306 crossrefType "posted-content" @default.
• W3005703306 hasAuthorship W3005703306A5040922600 @default.
• W3005703306 hasAuthorship W3005703306A5087313193 @default.
• W3005703306 hasConcept C11413529 @default.
• W3005703306 hasConcept C118615104 @default.
• W3005703306 hasConcept C122280245 @default.
• W3005703306 hasConcept C12267149 @default.
• W3005703306 hasConcept C126255220 @default.
• W3005703306 hasConcept C134306372 @default.
• W3005703306 hasConcept C134517425 @default.
• W3005703306 hasConcept C154945302 @default.
• W3005703306 hasConcept C162324750 @default.
• W3005703306 hasConcept C172623408 @default.
• W3005703306 hasConcept C195699287 @default.
• W3005703306 hasConcept C203616005 @default.
• W3005703306 hasConcept C2777303404 @default.
• W3005703306 hasConcept C28826006 @default.
• W3005703306 hasConcept C33923547 @default.
• W3005703306 hasConcept C41008148 @default.
• W3005703306 hasConcept C50522688 @default.
• W3005703306 hasConcept C62799726 @default.
• W3005703306 hasConcept C74193536 @default.
• W3005703306 hasConcept C80884492 @default.
• W3005703306 hasConceptScore W3005703306C11413529 @default.
• W3005703306 hasConceptScore W3005703306C118615104 @default.
• W3005703306 hasConceptScore W3005703306C122280245 @default.
• W3005703306 hasConceptScore W3005703306C12267149 @default.
• W3005703306 hasConceptScore W3005703306C126255220 @default.
• W3005703306 hasConceptScore W3005703306C134306372 @default.
• W3005703306 hasConceptScore W3005703306C134517425 @default.
• W3005703306 hasConceptScore W3005703306C154945302 @default.
• W3005703306 hasConceptScore W3005703306C162324750 @default.
• W3005703306 hasConceptScore W3005703306C172623408 @default.
• W3005703306 hasConceptScore W3005703306C195699287 @default.
• W3005703306 hasConceptScore W3005703306C203616005 @default.
• W3005703306 hasConceptScore W3005703306C2777303404 @default.
• W3005703306 hasConceptScore W3005703306C28826006 @default.
• W3005703306 hasConceptScore W3005703306C33923547 @default.
• W3005703306 hasConceptScore W3005703306C41008148 @default.
• W3005703306 hasConceptScore W3005703306C50522688 @default.
• W3005703306 hasConceptScore W3005703306C62799726 @default.
• W3005703306 hasConceptScore W3005703306C74193536 @default.
• W3005703306 hasConceptScore W3005703306C80884492 @default.
• W3005703306 hasLocation W30057033061 @default.
• W3005703306 hasOpenAccess W3005703306 @default.
• W3005703306 hasPrimaryLocation W30057033061 @default.
• W3005703306 hasRelatedWork W129562905 @default.
• W3005703306 hasRelatedWork W1605532809 @default.
• W3005703306 hasRelatedWork W2046047221 @default.
• W3005703306 hasRelatedWork W2108280195 @default.
• W3005703306 hasRelatedWork W2145759391 @default.

• next