FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3204183032> ?p ?o ?g. }

• W3204183032 endingPage "3869" @default.
• W3204183032 startingPage "3851" @default.
• W3204183032 abstract "Federated learning has emerged as a popular technique for distributing machine learning (ML) model training across the wireless edge. In this paper, we propose <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>two timescale hybrid federated learning</i> ( <monospace xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>TT-HF</monospace> ), a semi-decentralized learning architecture that combines the conventional device-to-server communication paradigm for federated learning with device-to-device (D2D) communications for model training. In <monospace xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>TT-HF</monospace> , during each global aggregation interval, devices (i) perform multiple stochastic gradient descent iterations on their individual datasets, and (ii) aperiodically engage in consensus procedure of their model parameters through cooperative, distributed D2D communications within local clusters. With a new general definition of gradient diversity, we formally study the convergence behavior of <monospace xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>TT-HF</monospace> , resulting in new convergence bounds for distributed ML. We leverage our convergence bounds to develop an adaptive control algorithm that tunes the step size, D2D communication rounds, and global aggregation period of <monospace xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>TT-HF</monospace> over time to target a sublinear convergence rate of <inline-formula xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink> <tex-math notation=LaTeX>$mathcal {O}(1/t)$ </tex-math></inline-formula> while minimizing network resource utilization. Our subsequent experiments demonstrate that <monospace xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>TT-HF</monospace> significantly outperforms the current art in federated learning in terms of model accuracy and/or network energy consumption in different scenarios where local device datasets exhibit statistical heterogeneity. Finally, our numerical evaluations demonstrate robustness against outages caused by fading channels, as well favorable performance with non-convex loss functions." @default.
• W3204183032 created "2021-10-11" @default.
• W3204183032 creator A5017461331 @default.
• W3204183032 creator A5020399355 @default.
• W3204183032 creator A5059750214 @default.
• W3204183032 creator A5064634430 @default.
• W3204183032 creator A5090464926 @default.
• W3204183032 date "2021-12-01" @default.
• W3204183032 modified "2023-10-15" @default.
• W3204183032 title "Semi-Decentralized Federated Learning With Cooperative D2D Local Model Aggregations" @default.
• W3204183032 cites W1677182931 @default.
• W3204183032 cites W1997834106 @default.
• W3204183032 cites W2074796812 @default.
• W3204183032 cites W2107338086 @default.
• W3204183032 cites W2123705108 @default.
• W3204183032 cites W2472333518 @default.
• W3204183032 cites W2740289948 @default.
• W3204183032 cites W2789254838 @default.
• W3204183032 cites W2912213068 @default.
• W3204183032 cites W2920095265 @default.
• W3204183032 cites W2963087201 @default.
• W3204183032 cites W2963136846 @default.
• W3204183032 cites W2963318081 @default.
• W3204183032 cites W2972093541 @default.
• W3204183032 cites W2975156709 @default.
• W3204183032 cites W2989289980 @default.
• W3204183032 cites W2996845627 @default.
• W3204183032 cites W2999074226 @default.
• W3204183032 cites W3008821091 @default.
• W3204183032 cites W3013888436 @default.
• W3204183032 cites W3015636663 @default.
• W3204183032 cites W3021096439 @default.
• W3204183032 cites W3045638580 @default.
• W3204183032 cites W3045747588 @default.
• W3204183032 cites W3046101409 @default.
• W3204183032 cites W3047079385 @default.
• W3204183032 cites W3088234149 @default.
• W3204183032 cites W3099980742 @default.
• W3204183032 cites W3100935401 @default.
• W3204183032 cites W3103657382 @default.
• W3204183032 cites W3103802018 @default.
• W3204183032 cites W3119842887 @default.
• W3204183032 cites W3120072747 @default.
• W3204183032 cites W3121122428 @default.
• W3204183032 cites W3123743074 @default.
• W3204183032 cites W3126228234 @default.
• W3204183032 cites W3134384882 @default.
• W3204183032 cites W3135231128 @default.
• W3204183032 cites W3155160971 @default.
• W3204183032 cites W3182131121 @default.
• W3204183032 cites W4206742934 @default.
• W3204183032 cites W4210984920 @default.
• W3204183032 doi "https://doi.org/10.1109/jsac.2021.3118344" @default.
• W3204183032 hasPublicationYear "2021" @default.
• W3204183032 type Work @default.
• W3204183032 sameAs 3204183032 @default.
• W3204183032 citedByCount "24" @default.
• W3204183032 countsByYear W32041830322021 @default.
• W3204183032 countsByYear W32041830322022 @default.
• W3204183032 countsByYear W32041830322023 @default.
• W3204183032 crossrefType "journal-article" @default.
• W3204183032 hasAuthorship W3204183032A5017461331 @default.
• W3204183032 hasAuthorship W3204183032A5020399355 @default.
• W3204183032 hasAuthorship W3204183032A5059750214 @default.
• W3204183032 hasAuthorship W3204183032A5064634430 @default.
• W3204183032 hasAuthorship W3204183032A5090464926 @default.
• W3204183032 hasBestOaLocation W32041830322 @default.
• W3204183032 hasConcept C11413529 @default.
• W3204183032 hasConcept C117160843 @default.
• W3204183032 hasConcept C118615104 @default.
• W3204183032 hasConcept C119857082 @default.
• W3204183032 hasConcept C153083717 @default.
• W3204183032 hasConcept C154945302 @default.
• W3204183032 hasConcept C162324750 @default.
• W3204183032 hasConcept C2777303404 @default.
• W3204183032 hasConcept C33923547 @default.
• W3204183032 hasConcept C41008148 @default.
• W3204183032 hasConcept C50522688 @default.
• W3204183032 hasConceptScore W3204183032C11413529 @default.
• W3204183032 hasConceptScore W3204183032C117160843 @default.
• W3204183032 hasConceptScore W3204183032C118615104 @default.
• W3204183032 hasConceptScore W3204183032C119857082 @default.
• W3204183032 hasConceptScore W3204183032C153083717 @default.
• W3204183032 hasConceptScore W3204183032C154945302 @default.
• W3204183032 hasConceptScore W3204183032C162324750 @default.
• W3204183032 hasConceptScore W3204183032C2777303404 @default.
• W3204183032 hasConceptScore W3204183032C33923547 @default.
• W3204183032 hasConceptScore W3204183032C41008148 @default.
• W3204183032 hasConceptScore W3204183032C50522688 @default.
• W3204183032 hasFunder F4320306076 @default.
• W3204183032 hasFunder F4320318238 @default.
• W3204183032 hasFunder F4320337345 @default.
• W3204183032 hasIssue "12" @default.
• W3204183032 hasLocation W32041830321 @default.
• W3204183032 hasLocation W32041830322 @default.
• W3204183032 hasOpenAccess W3204183032 @default.
• W3204183032 hasPrimaryLocation W32041830321 @default.
• W3204183032 hasRelatedWork W2961085424 @default.

• next