FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2896920867> ?p ?o ?g. }

• W2896920867 abstract "Valuable training data is often owned by independent organizations and located in multiple data centers. Most deep learning approaches require to centralize the multi-datacenter data for performance purpose. In practice, however, it is often infeasible to transfer all data to a centralized data center due to not only bandwidth limitation but also the constraints of privacy regulations. Model averaging is a conventional choice for data parallelized training, but its ineffectiveness is claimed by previous studies as deep neural networks are often non-convex. In this paper, we argue that model averaging can be effective in the decentralized environment by using two strategies, namely, the cyclical learning rate and the increased number of epochs for local model training. With the two strategies, we show that model averaging can provide competitive performance in the decentralized mode compared to the data-centralized one. In a practical environment with multiple data centers, we conduct extensive experiments using state-of-the-art deep network architectures on different types of data. Results demonstrate the effectiveness and robustness of the proposed method." @default.
• W2896920867 created "2018-10-26" @default.
• W2896920867 creator A5000582109 @default.
• W2896920867 creator A5010251502 @default.
• W2896920867 creator A5013340793 @default.
• W2896920867 creator A5018221313 @default.
• W2896920867 creator A5025503287 @default.
• W2896920867 creator A5039795290 @default.
• W2896920867 creator A5069084441 @default.
• W2896920867 date "2018-10-16" @default.
• W2896920867 modified "2023-09-23" @default.
• W2896920867 title "Collaborative Deep Learning Across Multiple Data Centers" @default.
• W2896920867 cites W1686810756 @default.
• W2896920867 cites W1779452081 @default.
• W2896920867 cites W1821462560 @default.
• W2896920867 cites W1850240193 @default.
• W2896920867 cites W2053637704 @default.
• W2896920867 cites W2095705004 @default.
• W2896920867 cites W2117539524 @default.
• W2896920867 cites W2166706236 @default.
• W2896920867 cites W2194775991 @default.
• W2896920867 cites W2283463896 @default.
• W2896920867 cites W2326433089 @default.
• W2896920867 cites W2407023693 @default.
• W2896920867 cites W2493916176 @default.
• W2896920867 cites W2541884796 @default.
• W2896920867 cites W2593116425 @default.
• W2896920867 cites W2604783387 @default.
• W2896920867 cites W2752711830 @default.
• W2896920867 cites W2785994986 @default.
• W2896920867 cites W2787017828 @default.
• W2896920867 cites W2791956393 @default.
• W2896920867 cites W2792287754 @default.
• W2896920867 cites W2949117887 @default.
• W2896920867 cites W2962758826 @default.
• W2896920867 cites W2963446712 @default.
• W2896920867 cites W2963539958 @default.
• W2896920867 cites W2963703618 @default.
• W2896920867 cites W2963803379 @default.
• W2896920867 cites W855165587 @default.
• W2896920867 doi "https://doi.org/10.48550/arxiv.1810.06877" @default.
• W2896920867 hasPublicationYear "2018" @default.
• W2896920867 type Work @default.
• W2896920867 sameAs 2896920867 @default.
• W2896920867 citedByCount "1" @default.
• W2896920867 countsByYear W28969208672021 @default.
• W2896920867 crossrefType "posted-content" @default.
• W2896920867 hasAuthorship W2896920867A5000582109 @default.
• W2896920867 hasAuthorship W2896920867A5010251502 @default.
• W2896920867 hasAuthorship W2896920867A5013340793 @default.
• W2896920867 hasAuthorship W2896920867A5018221313 @default.
• W2896920867 hasAuthorship W2896920867A5025503287 @default.
• W2896920867 hasAuthorship W2896920867A5039795290 @default.
• W2896920867 hasAuthorship W2896920867A5069084441 @default.
• W2896920867 hasBestOaLocation W28969208671 @default.
• W2896920867 hasConcept C104317684 @default.
• W2896920867 hasConcept C108583219 @default.
• W2896920867 hasConcept C119857082 @default.
• W2896920867 hasConcept C124101348 @default.
• W2896920867 hasConcept C150899416 @default.
• W2896920867 hasConcept C153740404 @default.
• W2896920867 hasConcept C154945302 @default.
• W2896920867 hasConcept C185592680 @default.
• W2896920867 hasConcept C2984842247 @default.
• W2896920867 hasConcept C31258907 @default.
• W2896920867 hasConcept C41008148 @default.
• W2896920867 hasConcept C50644808 @default.
• W2896920867 hasConcept C51632099 @default.
• W2896920867 hasConcept C55493867 @default.
• W2896920867 hasConcept C63479239 @default.
• W2896920867 hasConceptScore W2896920867C104317684 @default.
• W2896920867 hasConceptScore W2896920867C108583219 @default.
• W2896920867 hasConceptScore W2896920867C119857082 @default.
• W2896920867 hasConceptScore W2896920867C124101348 @default.
• W2896920867 hasConceptScore W2896920867C150899416 @default.
• W2896920867 hasConceptScore W2896920867C153740404 @default.
• W2896920867 hasConceptScore W2896920867C154945302 @default.
• W2896920867 hasConceptScore W2896920867C185592680 @default.
• W2896920867 hasConceptScore W2896920867C2984842247 @default.
• W2896920867 hasConceptScore W2896920867C31258907 @default.
• W2896920867 hasConceptScore W2896920867C41008148 @default.
• W2896920867 hasConceptScore W2896920867C50644808 @default.
• W2896920867 hasConceptScore W2896920867C51632099 @default.
• W2896920867 hasConceptScore W2896920867C55493867 @default.
• W2896920867 hasConceptScore W2896920867C63479239 @default.
• W2896920867 hasLocation W28969208671 @default.
• W2896920867 hasLocation W28969208672 @default.
• W2896920867 hasOpenAccess W2896920867 @default.
• W2896920867 hasPrimaryLocation W28969208671 @default.
• W2896920867 hasRelatedWork W2889705046 @default.
• W2896920867 hasRelatedWork W2946016983 @default.
• W2896920867 hasRelatedWork W2960456850 @default.
• W2896920867 hasRelatedWork W2980128178 @default.
• W2896920867 hasRelatedWork W4213250359 @default.
• W2896920867 hasRelatedWork W4288040045 @default.
• W2896920867 hasRelatedWork W4312200629 @default.
• W2896920867 hasRelatedWork W4317565044 @default.
• W2896920867 hasRelatedWork W4382286161 @default.
• W2896920867 hasRelatedWork W4386213806 @default.
• W2896920867 isParatext "false" @default.

• next