Matches in SemOpenAlex for { <https://semopenalex.org/work/W2951670102> ?p ?o ?g. }
- W2951670102 abstract "A new amortized variance-reduced gradient (AVRG) algorithm was developed in cite{ying2017convergence}, which has constant storage requirement in comparison to SAGA and balanced gradient computations in comparison to SVRG. One key advantage of the AVRG strategy is its amenability to decentralized implementations. In this work, we show how AVRG can be extended to the network case where multiple learning agents are assumed to be connected by a graph topology. In this scenario, each agent observes data that is spatially distributed and all agents are only allowed to communicate with direct neighbors. Moreover, the amount of data observed by the individual agents may differ drastically. For such situations, the balanced gradient computation property of AVRG becomes a real advantage in reducing idle time caused by unbalanced local data storage requirements, which is characteristic of other reduced-variance gradient algorithms. The resulting diffusion-AVRG algorithm is shown to have linear convergence to the exact solution, and is much more memory efficient than other alternative algorithms. In addition, we propose a mini-batch strategy to balance the communication and computation efficiency for diffusion-AVRG. When a proper batch size is employed, it is observed in simulations that diffusion-AVRG is more computationally efficient than exact diffusion or EXTRA while maintaining almost the same communication efficiency." @default.
- W2951670102 created "2019-06-27" @default.
- W2951670102 creator A5005538082 @default.
- W2951670102 creator A5007519257 @default.
- W2951670102 creator A5084663182 @default.
- W2951670102 creator A5091103447 @default.
- W2951670102 date "2017-08-04" @default.
- W2951670102 modified "2023-10-16" @default.
- W2951670102 title "Variance-Reduced Stochastic Learning by Networked Agents under Random Reshuffling" @default.
- W2951670102 cites W114517082 @default.
- W2951670102 cites W1553365534 @default.
- W2951670102 cites W1571416372 @default.
- W2951670102 cites W1602732414 @default.
- W2951670102 cites W1616857247 @default.
- W2951670102 cites W1627780903 @default.
- W2951670102 cites W2012445782 @default.
- W2951670102 cites W2027481851 @default.
- W2951670102 cites W2047152541 @default.
- W2951670102 cites W2068254488 @default.
- W2951670102 cites W2107438106 @default.
- W2951670102 cites W2108306501 @default.
- W2951670102 cites W2123705108 @default.
- W2951670102 cites W2216483973 @default.
- W2951670102 cites W2235185944 @default.
- W2951670102 cites W2253399765 @default.
- W2951670102 cites W2291434103 @default.
- W2951670102 cites W2295500362 @default.
- W2951670102 cites W2530417694 @default.
- W2951670102 cites W2612249769 @default.
- W2951670102 cites W2742134852 @default.
- W2951670102 cites W2951224092 @default.
- W2951670102 cites W2951841641 @default.
- W2951670102 cites W2952215077 @default.
- W2951670102 cites W2963649943 @default.
- W2951670102 cites W2963992805 @default.
- W2951670102 cites W3104550706 @default.
- W2951670102 doi "https://doi.org/10.48550/arxiv.1708.01384" @default.
- W2951670102 hasPublicationYear "2017" @default.
- W2951670102 type Work @default.
- W2951670102 sameAs 2951670102 @default.
- W2951670102 citedByCount "1" @default.
- W2951670102 countsByYear W29516701022017 @default.
- W2951670102 crossrefType "posted-content" @default.
- W2951670102 hasAuthorship W2951670102A5005538082 @default.
- W2951670102 hasAuthorship W2951670102A5007519257 @default.
- W2951670102 hasAuthorship W2951670102A5084663182 @default.
- W2951670102 hasAuthorship W2951670102A5091103447 @default.
- W2951670102 hasBestOaLocation W29516701021 @default.
- W2951670102 hasConcept C11413529 @default.
- W2951670102 hasConcept C114614502 @default.
- W2951670102 hasConcept C121332964 @default.
- W2951670102 hasConcept C121955636 @default.
- W2951670102 hasConcept C126255220 @default.
- W2951670102 hasConcept C144133560 @default.
- W2951670102 hasConcept C162324750 @default.
- W2951670102 hasConcept C184720557 @default.
- W2951670102 hasConcept C196083921 @default.
- W2951670102 hasConcept C199360897 @default.
- W2951670102 hasConcept C26713055 @default.
- W2951670102 hasConcept C2777303404 @default.
- W2951670102 hasConcept C33923547 @default.
- W2951670102 hasConcept C41008148 @default.
- W2951670102 hasConcept C45374587 @default.
- W2951670102 hasConcept C50522688 @default.
- W2951670102 hasConcept C69357855 @default.
- W2951670102 hasConcept C97355855 @default.
- W2951670102 hasConceptScore W2951670102C11413529 @default.
- W2951670102 hasConceptScore W2951670102C114614502 @default.
- W2951670102 hasConceptScore W2951670102C121332964 @default.
- W2951670102 hasConceptScore W2951670102C121955636 @default.
- W2951670102 hasConceptScore W2951670102C126255220 @default.
- W2951670102 hasConceptScore W2951670102C144133560 @default.
- W2951670102 hasConceptScore W2951670102C162324750 @default.
- W2951670102 hasConceptScore W2951670102C184720557 @default.
- W2951670102 hasConceptScore W2951670102C196083921 @default.
- W2951670102 hasConceptScore W2951670102C199360897 @default.
- W2951670102 hasConceptScore W2951670102C26713055 @default.
- W2951670102 hasConceptScore W2951670102C2777303404 @default.
- W2951670102 hasConceptScore W2951670102C33923547 @default.
- W2951670102 hasConceptScore W2951670102C41008148 @default.
- W2951670102 hasConceptScore W2951670102C45374587 @default.
- W2951670102 hasConceptScore W2951670102C50522688 @default.
- W2951670102 hasConceptScore W2951670102C69357855 @default.
- W2951670102 hasConceptScore W2951670102C97355855 @default.
- W2951670102 hasLocation W29516701021 @default.
- W2951670102 hasOpenAccess W2951670102 @default.
- W2951670102 hasPrimaryLocation W29516701021 @default.
- W2951670102 hasRelatedWork W1485630101 @default.
- W2951670102 hasRelatedWork W1504651774 @default.
- W2951670102 hasRelatedWork W1536235319 @default.
- W2951670102 hasRelatedWork W1575263621 @default.
- W2951670102 hasRelatedWork W1964111820 @default.
- W2951670102 hasRelatedWork W2083338789 @default.
- W2951670102 hasRelatedWork W2354062721 @default.
- W2951670102 hasRelatedWork W2384497200 @default.
- W2951670102 hasRelatedWork W4244853187 @default.
- W2951670102 hasRelatedWork W2803721528 @default.
- W2951670102 isParatext "false" @default.
- W2951670102 isRetracted "false" @default.
- W2951670102 magId "2951670102" @default.