Matches in SemOpenAlex for { <https://semopenalex.org/work/W4387544239> ?p ?o ?g. }
Showing items 1 to 71 of
71
with 100 items per page.
- W4387544239 abstract "Gradient methods, such as gradient descent, are widely deployed to train optimization-based models in machine learning. To train such models on a large dataset, the dataset is commonly split into multiple partitions which are trained on different workers. In order to tolerate stragglers, existing techniques either use gradient coding (GC) to recover the full gradients from a certain number of workers or recover gradients partially from an arbitrary number of workers. In this paper, we propose ignore-straggler gradient coding (IS-GC) that allows GC to tolerate an arbitrary number of stragglers. Compared to approximated gradient descent, IS-GC can recover more gradients when there is the same number of stragglers. We design a graph-based model to decode coded gradients from an arbitrary number of stragglers, and prove that it can maximize the recovery of gradients. We apply IS-GC on fractional repetition (FR) and cyclic repetition (CR), two representative dataset placement schemes of GC. We also propose hybrid repetition (HR) that generalizes over FR and CR and achieves a flexible trade-off between FR and CR. With extensive experiments, we demonstrate that IS-GC can flexibly tolerate an arbitrary number of stragglers and achieve a low completion time of training." @default.
- W4387544239 created "2023-10-12" @default.
- W4387544239 creator A5028880618 @default.
- W4387544239 creator A5077790779 @default.
- W4387544239 creator A5079337517 @default.
- W4387544239 date "2023-07-01" @default.
- W4387544239 modified "2023-10-12" @default.
- W4387544239 title "On Arbitrary Ignorance of Stragglers with Gradient Coding" @default.
- W4387544239 cites W1497601926 @default.
- W4387544239 cites W1582786162 @default.
- W4387544239 cites W1963707128 @default.
- W4387544239 cites W2001258629 @default.
- W4387544239 cites W2009990213 @default.
- W4387544239 cites W2125993116 @default.
- W4387544239 cites W2268702383 @default.
- W4387544239 cites W2523435939 @default.
- W4387544239 cites W2913243214 @default.
- W4387544239 cites W2962208489 @default.
- W4387544239 cites W2963706835 @default.
- W4387544239 cites W2964323106 @default.
- W4387544239 cites W3016249977 @default.
- W4387544239 cites W3023838981 @default.
- W4387544239 cites W3098226281 @default.
- W4387544239 cites W3197511597 @default.
- W4387544239 cites W4234503811 @default.
- W4387544239 doi "https://doi.org/10.1109/icdcs57875.2023.00080" @default.
- W4387544239 hasPublicationYear "2023" @default.
- W4387544239 type Work @default.
- W4387544239 citedByCount "0" @default.
- W4387544239 crossrefType "proceedings-article" @default.
- W4387544239 hasAuthorship W4387544239A5028880618 @default.
- W4387544239 hasAuthorship W4387544239A5077790779 @default.
- W4387544239 hasAuthorship W4387544239A5079337517 @default.
- W4387544239 hasConcept C105795698 @default.
- W4387544239 hasConcept C11413529 @default.
- W4387544239 hasConcept C138885662 @default.
- W4387544239 hasConcept C153258448 @default.
- W4387544239 hasConcept C154945302 @default.
- W4387544239 hasConcept C179518139 @default.
- W4387544239 hasConcept C2776141515 @default.
- W4387544239 hasConcept C33923547 @default.
- W4387544239 hasConcept C41008148 @default.
- W4387544239 hasConcept C41895202 @default.
- W4387544239 hasConcept C50644808 @default.
- W4387544239 hasConceptScore W4387544239C105795698 @default.
- W4387544239 hasConceptScore W4387544239C11413529 @default.
- W4387544239 hasConceptScore W4387544239C138885662 @default.
- W4387544239 hasConceptScore W4387544239C153258448 @default.
- W4387544239 hasConceptScore W4387544239C154945302 @default.
- W4387544239 hasConceptScore W4387544239C179518139 @default.
- W4387544239 hasConceptScore W4387544239C2776141515 @default.
- W4387544239 hasConceptScore W4387544239C33923547 @default.
- W4387544239 hasConceptScore W4387544239C41008148 @default.
- W4387544239 hasConceptScore W4387544239C41895202 @default.
- W4387544239 hasConceptScore W4387544239C50644808 @default.
- W4387544239 hasLocation W43875442391 @default.
- W4387544239 hasOpenAccess W4387544239 @default.
- W4387544239 hasPrimaryLocation W43875442391 @default.
- W4387544239 hasRelatedWork W11343509 @default.
- W4387544239 hasRelatedWork W2003474770 @default.
- W4387544239 hasRelatedWork W2351132524 @default.
- W4387544239 hasRelatedWork W2392934913 @default.
- W4387544239 hasRelatedWork W2612358220 @default.
- W4387544239 hasRelatedWork W2748952813 @default.
- W4387544239 hasRelatedWork W2899084033 @default.
- W4387544239 hasRelatedWork W2916738897 @default.
- W4387544239 hasRelatedWork W2996195527 @default.
- W4387544239 hasRelatedWork W3011277146 @default.
- W4387544239 isParatext "false" @default.
- W4387544239 isRetracted "false" @default.
- W4387544239 workType "article" @default.