FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2891800530> ?p ?o ?g. }

• W2891800530 abstract "Modern distributed training of machine learning models suffers from high communication overhead for synchronizing stochastic gradients and model parameters. In this paper, to reduce the communication complexity, we propose emph{double quantization}, a general scheme for quantizing both model parameters and gradients. Three communication-efficient algorithms are proposed under this general scheme. Specifically, (i) we propose a low-precision algorithm AsyLPG with asynchronous parallelism, (ii) we explore integrating gradient sparsification with double quantization and develop Sparse-AsyLPG, (iii) we show that double quantization can also be accelerated by momentum technique and design accelerated AsyLPG. We establish rigorous performance guarantees for the algorithms, and conduct experiments on a multi-server test-bed to demonstrate that our algorithms can effectively save transmitted bits without performance degradation." @default.
• W2891800530 created "2018-09-27" @default.
• W2891800530 creator A5000295734 @default.
• W2891800530 creator A5042209817 @default.
• W2891800530 creator A5082905458 @default.
• W2891800530 date "2018-05-25" @default.
• W2891800530 modified "2023-09-24" @default.
• W2891800530 title "Double Quantization for Communication-Efficient Distributed Optimization" @default.
• W2891800530 cites W1501362825 @default.
• W2891800530 cites W1843828865 @default.
• W2891800530 cites W1845277745 @default.
• W2891800530 cites W2009537245 @default.
• W2891800530 cites W2024484010 @default.
• W2891800530 cites W2029463628 @default.
• W2891800530 cites W2038497950 @default.
• W2891800530 cites W2044212084 @default.
• W2891800530 cites W2049469158 @default.
• W2891800530 cites W2083842231 @default.
• W2891800530 cites W2103628399 @default.
• W2891800530 cites W2107438106 @default.
• W2891800530 cites W2119200797 @default.
• W2891800530 cites W2138243089 @default.
• W2891800530 cites W2153635508 @default.
• W2891800530 cites W2162361955 @default.
• W2891800530 cites W2167732364 @default.
• W2891800530 cites W2168231600 @default.
• W2891800530 cites W2205410708 @default.
• W2891800530 cites W2342177834 @default.
• W2891800530 cites W2407022425 @default.
• W2891800530 cites W2407904812 @default.
• W2891800530 cites W2408675156 @default.
• W2891800530 cites W2414455940 @default.
• W2891800530 cites W2432353846 @default.
• W2891800530 cites W2519604671 @default.
• W2891800530 cites W2525160011 @default.
• W2891800530 cites W2569406209 @default.
• W2891800530 cites W2593002563 @default.
• W2891800530 cites W2617766261 @default.
• W2891800530 cites W2647836899 @default.
• W2891800530 cites W2741269719 @default.
• W2891800530 cites W2787950853 @default.
• W2891800530 cites W2798899619 @default.
• W2891800530 cites W2803965032 @default.
• W2891800530 cites W2950948235 @default.
• W2891800530 cites W2951078008 @default.
• W2891800530 cites W2951781666 @default.
• W2891800530 cites W2962890301 @default.
• W2891800530 cites W2963903325 @default.
• W2891800530 cites W3141595720 @default.
• W2891800530 hasPublicationYear "2018" @default.
• W2891800530 type Work @default.
• W2891800530 sameAs 2891800530 @default.
• W2891800530 citedByCount "1" @default.
• W2891800530 countsByYear W28918005302021 @default.
• W2891800530 crossrefType "posted-content" @default.
• W2891800530 hasAuthorship W2891800530A5000295734 @default.
• W2891800530 hasAuthorship W2891800530A5042209817 @default.
• W2891800530 hasAuthorship W2891800530A5082905458 @default.
• W2891800530 hasConcept C113775141 @default.
• W2891800530 hasConcept C11413529 @default.
• W2891800530 hasConcept C134306372 @default.
• W2891800530 hasConcept C151319957 @default.
• W2891800530 hasConcept C162932704 @default.
• W2891800530 hasConcept C28855332 @default.
• W2891800530 hasConcept C31258907 @default.
• W2891800530 hasConcept C33923547 @default.
• W2891800530 hasConcept C41008148 @default.
• W2891800530 hasConcept C761482 @default.
• W2891800530 hasConcept C76155785 @default.
• W2891800530 hasConcept C77618280 @default.
• W2891800530 hasConceptScore W2891800530C113775141 @default.
• W2891800530 hasConceptScore W2891800530C11413529 @default.
• W2891800530 hasConceptScore W2891800530C134306372 @default.
• W2891800530 hasConceptScore W2891800530C151319957 @default.
• W2891800530 hasConceptScore W2891800530C162932704 @default.
• W2891800530 hasConceptScore W2891800530C28855332 @default.
• W2891800530 hasConceptScore W2891800530C31258907 @default.
• W2891800530 hasConceptScore W2891800530C33923547 @default.
• W2891800530 hasConceptScore W2891800530C41008148 @default.
• W2891800530 hasConceptScore W2891800530C761482 @default.
• W2891800530 hasConceptScore W2891800530C76155785 @default.
• W2891800530 hasConceptScore W2891800530C77618280 @default.
• W2891800530 hasLocation W28918005301 @default.
• W2891800530 hasOpenAccess W2891800530 @default.
• W2891800530 hasPrimaryLocation W28918005301 @default.
• W2891800530 hasRelatedWork W2112159083 @default.
• W2891800530 hasRelatedWork W2497335382 @default.
• W2891800530 hasRelatedWork W2804635205 @default.
• W2891800530 hasRelatedWork W2891735383 @default.
• W2891800530 hasRelatedWork W2892799969 @default.
• W2891800530 hasRelatedWork W2911338751 @default.
• W2891800530 hasRelatedWork W2964163156 @default.
• W2891800530 hasRelatedWork W2970472742 @default.
• W2891800530 hasRelatedWork W3006817794 @default.
• W2891800530 hasRelatedWork W3016081801 @default.
• W2891800530 hasRelatedWork W3030474951 @default.
• W2891800530 hasRelatedWork W3109509664 @default.
• W2891800530 hasRelatedWork W3113797829 @default.
• W2891800530 hasRelatedWork W3123666279 @default.
• W2891800530 hasRelatedWork W3135784973 @default.

• next