Matches in SemOpenAlex for { <https://semopenalex.org/work/W2913310393> ?p ?o ?g. }
- W2913310393 abstract "We present two sampled quasi-Newton methods for deep learning: sampled LBFGS (S-LBFGS) and sampled LSR1 (S-LSR1). Contrary to the classical variants of these methods that sequentially build Hessian or inverse Hessian approximations as the optimization progresses, our proposed methods sample points randomly around the current iterate at every iteration to produce these approximations. As a result, the approximations constructed make use of more reliable (recent and local) information, and do not depend on past iterate information that could be significantly stale. Our proposed algorithms are efficient in terms of accessed data points (epochs) and have enough concurrency to take advantage of parallel/distributed computing environments. We provide convergence guarantees for our proposed methods. Numerical tests on a toy classification problem as well as on popular benchmarking neural network training tasks reveal that the methods outperform their classical variants." @default.
- W2913310393 created "2019-02-21" @default.
- W2913310393 creator A5029030164 @default.
- W2913310393 creator A5064222013 @default.
- W2913310393 creator A5070679093 @default.
- W2913310393 date "2019-01-28" @default.
- W2913310393 modified "2023-09-27" @default.
- W2913310393 title "Quasi-Newton Methods for Deep Learning: Forget the Past, Just Sample." @default.
- W2913310393 cites W147998453 @default.
- W2913310393 cites W1491622225 @default.
- W2913310393 cites W1522301498 @default.
- W2913310393 cites W1592294486 @default.
- W2913310393 cites W1780115997 @default.
- W2913310393 cites W1966150378 @default.
- W2913310393 cites W196761320 @default.
- W2913310393 cites W1991083751 @default.
- W2913310393 cites W1994616650 @default.
- W2913310393 cites W1998749910 @default.
- W2913310393 cites W2005136695 @default.
- W2913310393 cites W2025469203 @default.
- W2913310393 cites W2026393324 @default.
- W2913310393 cites W2038210983 @default.
- W2913310393 cites W2051434435 @default.
- W2913310393 cites W2051669046 @default.
- W2913310393 cites W2053964895 @default.
- W2913310393 cites W2058839679 @default.
- W2913310393 cites W2064217481 @default.
- W2913310393 cites W2078409719 @default.
- W2913310393 cites W2079482358 @default.
- W2913310393 cites W2092087339 @default.
- W2913310393 cites W2095211439 @default.
- W2913310393 cites W2102486516 @default.
- W2913310393 cites W2107438106 @default.
- W2913310393 cites W2112796928 @default.
- W2913310393 cites W2146502635 @default.
- W2913310393 cites W2154834860 @default.
- W2913310393 cites W2287011250 @default.
- W2913310393 cites W2404678399 @default.
- W2913310393 cites W2442014075 @default.
- W2913310393 cites W2523060838 @default.
- W2913310393 cites W2594182398 @default.
- W2913310393 cites W2622263826 @default.
- W2913310393 cites W2747280560 @default.
- W2913310393 cites W2769856846 @default.
- W2913310393 cites W2793117408 @default.
- W2913310393 cites W2884711234 @default.
- W2913310393 cites W2898280316 @default.
- W2913310393 cites W2900789157 @default.
- W2913310393 cites W2951781666 @default.
- W2913310393 cites W2952215077 @default.
- W2913310393 cites W2962747323 @default.
- W2913310393 cites W2962986403 @default.
- W2913310393 cites W2963160732 @default.
- W2913310393 cites W2963242954 @default.
- W2913310393 cites W2963307318 @default.
- W2913310393 cites W2963397933 @default.
- W2913310393 cites W2963433607 @default.
- W2913310393 cites W2963794891 @default.
- W2913310393 cites W2963813563 @default.
- W2913310393 cites W2963941964 @default.
- W2913310393 cites W2964312150 @default.
- W2913310393 cites W2970227300 @default.
- W2913310393 cites W3005768517 @default.
- W2913310393 cites W3029645440 @default.
- W2913310393 cites W3030916542 @default.
- W2913310393 cites W3118608800 @default.
- W2913310393 cites W3141890267 @default.
- W2913310393 cites W778657980 @default.
- W2913310393 hasPublicationYear "2019" @default.
- W2913310393 type Work @default.
- W2913310393 sameAs 2913310393 @default.
- W2913310393 citedByCount "19" @default.
- W2913310393 countsByYear W29133103932018 @default.
- W2913310393 countsByYear W29133103932019 @default.
- W2913310393 countsByYear W29133103932020 @default.
- W2913310393 countsByYear W29133103932021 @default.
- W2913310393 crossrefType "posted-content" @default.
- W2913310393 hasAuthorship W2913310393A5029030164 @default.
- W2913310393 hasAuthorship W2913310393A5064222013 @default.
- W2913310393 hasAuthorship W2913310393A5070679093 @default.
- W2913310393 hasConcept C108583219 @default.
- W2913310393 hasConcept C111919701 @default.
- W2913310393 hasConcept C11413529 @default.
- W2913310393 hasConcept C114954040 @default.
- W2913310393 hasConcept C119857082 @default.
- W2913310393 hasConcept C121332964 @default.
- W2913310393 hasConcept C126255220 @default.
- W2913310393 hasConcept C144133560 @default.
- W2913310393 hasConcept C154945302 @default.
- W2913310393 hasConcept C158622935 @default.
- W2913310393 hasConcept C162324750 @default.
- W2913310393 hasConcept C162853370 @default.
- W2913310393 hasConcept C185592680 @default.
- W2913310393 hasConcept C193702766 @default.
- W2913310393 hasConcept C198531522 @default.
- W2913310393 hasConcept C203616005 @default.
- W2913310393 hasConcept C207467116 @default.
- W2913310393 hasConcept C2524010 @default.
- W2913310393 hasConcept C2777303404 @default.
- W2913310393 hasConcept C28826006 @default.