FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3083746972> ?p ?o ?g. }

• W3083746972 abstract "Lossy gradient compression, with either unbiased or biased compressors, has become a key tool to avoid the communication bottleneck in centrally coordinated distributed training of machine learning models. We analyze the performance of two standard and general types of methods: (i) distributed quantized SGD (D-QSGD) with arbitrary unbiased quantizers and (ii) distributed SGD with error-feedback and biased compressors (D-EF-SGD) in the heterogeneous (non-iid) data setting. Our results indicate that D-EF-SGD is much less affected than D-QSGD by non-iid data, but both methods can suffer a slowdown if data-skewness is high. We further study two alternatives that are not (or much less) affected by heterogenous data distributions: first, a recently proposed method that is effective on strongly convex problems, and secondly, we point out a more general approach that is applicable to linear compressors only but effective in all considered scenarios." @default.
• W3083746972 created "2020-09-11" @default.
• W3083746972 creator A5045519627 @default.
• W3083746972 date "2020-09-04" @default.
• W3083746972 modified "2023-10-01" @default.
• W3083746972 title "On Communication Compression for Distributed Optimization on Heterogeneous Data." @default.
• W3083746972 cites W114517082 @default.
• W3083746972 cites W2146502635 @default.
• W3083746972 cites W2405578611 @default.
• W3083746972 cites W2407022425 @default.
• W3083746972 cites W2535838896 @default.
• W3083746972 cites W2741269719 @default.
• W3083746972 cites W2769644379 @default.
• W3083746972 cites W2785225676 @default.
• W3083746972 cites W2808495300 @default.
• W3083746972 cites W2889676205 @default.
• W3083746972 cites W2890924858 @default.
• W3083746972 cites W2911327153 @default.
• W3083746972 cites W2911863041 @default.
• W3083746972 cites W2913259183 @default.
• W3083746972 cites W2914138317 @default.
• W3083746972 cites W2935990182 @default.
• W3083746972 cites W2948746540 @default.
• W3083746972 cites W2958992393 @default.
• W3083746972 cites W2963433607 @default.
• W3083746972 cites W2963540381 @default.
• W3083746972 cites W2963664311 @default.
• W3083746972 cites W2963766684 @default.
• W3083746972 cites W2963803379 @default.
• W3083746972 cites W2964004663 @default.
• W3083746972 cites W2964121744 @default.
• W3083746972 cites W2965234876 @default.
• W3083746972 cites W2970289928 @default.
• W3083746972 cites W2970421227 @default.
• W3083746972 cites W2973096819 @default.
• W3083746972 cites W2981206218 @default.
• W3083746972 cites W2995653155 @default.
• W3083746972 cites W2996118901 @default.
• W3083746972 cites W3007528421 @default.
• W3083746972 cites W3034713518 @default.
• W3083746972 cites W3034741284 @default.
• W3083746972 cites W3036623140 @default.
• W3083746972 cites W3043574444 @default.
• W3083746972 cites W3047411835 @default.
• W3083746972 cites W3100770245 @default.
• W3083746972 cites W3101036738 @default.
• W3083746972 cites W3102089081 @default.
• W3083746972 cites W584312462 @default.
• W3083746972 hasPublicationYear "2020" @default.
• W3083746972 type Work @default.
• W3083746972 sameAs 3083746972 @default.
• W3083746972 citedByCount "4" @default.
• W3083746972 countsByYear W30837469722019 @default.
• W3083746972 countsByYear W30837469722021 @default.
• W3083746972 countsByYear W30837469722022 @default.
• W3083746972 crossrefType "posted-content" @default.
• W3083746972 hasAuthorship W3083746972A5045519627 @default.
• W3083746972 hasConcept C112680207 @default.
• W3083746972 hasConcept C11413529 @default.
• W3083746972 hasConcept C120314980 @default.
• W3083746972 hasConcept C122342681 @default.
• W3083746972 hasConcept C149635348 @default.
• W3083746972 hasConcept C149782125 @default.
• W3083746972 hasConcept C154945302 @default.
• W3083746972 hasConcept C159985019 @default.
• W3083746972 hasConcept C165021410 @default.
• W3083746972 hasConcept C180016635 @default.
• W3083746972 hasConcept C192562407 @default.
• W3083746972 hasConcept C2524010 @default.
• W3083746972 hasConcept C26517878 @default.
• W3083746972 hasConcept C2780513914 @default.
• W3083746972 hasConcept C33923547 @default.
• W3083746972 hasConcept C38652104 @default.
• W3083746972 hasConcept C41008148 @default.
• W3083746972 hasConcept C43711488 @default.
• W3083746972 hasConcept C76155785 @default.
• W3083746972 hasConcept C78548338 @default.
• W3083746972 hasConceptScore W3083746972C112680207 @default.
• W3083746972 hasConceptScore W3083746972C11413529 @default.
• W3083746972 hasConceptScore W3083746972C120314980 @default.
• W3083746972 hasConceptScore W3083746972C122342681 @default.
• W3083746972 hasConceptScore W3083746972C149635348 @default.
• W3083746972 hasConceptScore W3083746972C149782125 @default.
• W3083746972 hasConceptScore W3083746972C154945302 @default.
• W3083746972 hasConceptScore W3083746972C159985019 @default.
• W3083746972 hasConceptScore W3083746972C165021410 @default.
• W3083746972 hasConceptScore W3083746972C180016635 @default.
• W3083746972 hasConceptScore W3083746972C192562407 @default.
• W3083746972 hasConceptScore W3083746972C2524010 @default.
• W3083746972 hasConceptScore W3083746972C26517878 @default.
• W3083746972 hasConceptScore W3083746972C2780513914 @default.
• W3083746972 hasConceptScore W3083746972C33923547 @default.
• W3083746972 hasConceptScore W3083746972C38652104 @default.
• W3083746972 hasConceptScore W3083746972C41008148 @default.
• W3083746972 hasConceptScore W3083746972C43711488 @default.
• W3083746972 hasConceptScore W3083746972C76155785 @default.
• W3083746972 hasConceptScore W3083746972C78548338 @default.
• W3083746972 hasLocation W30837469721 @default.
• W3083746972 hasOpenAccess W3083746972 @default.
• W3083746972 hasPrimaryLocation W30837469721 @default.

• next