Matches in SemOpenAlex for { <https://semopenalex.org/work/W3095445260> ?p ?o ?g. }
- W3095445260 abstract "Decentralized optimization methods enable on-device training of machine learning models without a central coordinator. In many scenarios communication between devices is energy demanding and time consuming and forms the bottleneck of the entire system. We propose a new randomized first-order method which tackles the communication bottleneck by applying randomized compression operators to the communicated messages. By combining our scheme with a new variance reduction technique that progressively throughout the iterations reduces the adverse effect of the injected quantization noise, we obtain the first scheme that converges linearly on strongly convex decentralized problems while using compressed communication only. We prove that our method can solve the problems without any increase in the number of communications compared to the baseline which does not perform any communication compression while still allowing for a significant compression factor which depends on the conditioning of the problem and the topology of the network. Our key theoretical findings are supported by numerical experiments." @default.
- W3095445260 created "2020-11-09" @default.
- W3095445260 creator A5017429728 @default.
- W3095445260 creator A5036598221 @default.
- W3095445260 creator A5045519627 @default.
- W3095445260 creator A5065163339 @default.
- W3095445260 creator A5073756389 @default.
- W3095445260 date "2020-11-03" @default.
- W3095445260 modified "2023-10-01" @default.
- W3095445260 title "A Linearly Convergent Algorithm for Decentralized Optimization: Sending Less Bits for Free!" @default.
- W3095445260 cites W1507042982 @default.
- W3095445260 cites W1512098439 @default.
- W3095445260 cites W1547358136 @default.
- W3095445260 cites W1571416372 @default.
- W3095445260 cites W1616857247 @default.
- W3095445260 cites W1980819180 @default.
- W3095445260 cites W2024175297 @default.
- W3095445260 cites W2028201344 @default.
- W3095445260 cites W2044212084 @default.
- W3095445260 cites W2045079045 @default.
- W3095445260 cites W2046387702 @default.
- W3095445260 cites W2047894825 @default.
- W3095445260 cites W2080844831 @default.
- W3095445260 cites W2098863504 @default.
- W3095445260 cites W2100403032 @default.
- W3095445260 cites W2107438106 @default.
- W3095445260 cites W2107664305 @default.
- W3095445260 cites W2123705108 @default.
- W3095445260 cites W2135482703 @default.
- W3095445260 cites W2150704725 @default.
- W3095445260 cites W2164278908 @default.
- W3095445260 cites W2168231600 @default.
- W3095445260 cites W2216483973 @default.
- W3095445260 cites W2243169878 @default.
- W3095445260 cites W2283463896 @default.
- W3095445260 cites W2400034325 @default.
- W3095445260 cites W2405578611 @default.
- W3095445260 cites W2407022425 @default.
- W3095445260 cites W2513383847 @default.
- W3095445260 cites W2541884796 @default.
- W3095445260 cites W2547352193 @default.
- W3095445260 cites W2591577911 @default.
- W3095445260 cites W2889676205 @default.
- W3095445260 cites W2890924858 @default.
- W3095445260 cites W2892209803 @default.
- W3095445260 cites W2911327153 @default.
- W3095445260 cites W2911863041 @default.
- W3095445260 cites W2935990182 @default.
- W3095445260 cites W2962915227 @default.
- W3095445260 cites W2963228337 @default.
- W3095445260 cites W2963357609 @default.
- W3095445260 cites W2963540381 @default.
- W3095445260 cites W2963617134 @default.
- W3095445260 cites W2963649943 @default.
- W3095445260 cites W2963855576 @default.
- W3095445260 cites W2964004663 @default.
- W3095445260 cites W2965142130 @default.
- W3095445260 cites W2965234876 @default.
- W3095445260 cites W2973096819 @default.
- W3095445260 cites W2991772480 @default.
- W3095445260 cites W2995022099 @default.
- W3095445260 cites W2996118901 @default.
- W3095445260 cites W3011252985 @default.
- W3095445260 cites W3016897523 @default.
- W3095445260 cites W3022064003 @default.
- W3095445260 cites W3023944502 @default.
- W3095445260 cites W3027921946 @default.
- W3095445260 cites W3034741284 @default.
- W3095445260 cites W3130135650 @default.
- W3095445260 cites W584312462 @default.
- W3095445260 hasPublicationYear "2020" @default.
- W3095445260 type Work @default.
- W3095445260 sameAs 3095445260 @default.
- W3095445260 citedByCount "3" @default.
- W3095445260 countsByYear W30954452602020 @default.
- W3095445260 countsByYear W30954452602021 @default.
- W3095445260 crossrefType "posted-content" @default.
- W3095445260 hasAuthorship W3095445260A5017429728 @default.
- W3095445260 hasAuthorship W3095445260A5036598221 @default.
- W3095445260 hasAuthorship W3095445260A5045519627 @default.
- W3095445260 hasAuthorship W3095445260A5065163339 @default.
- W3095445260 hasAuthorship W3095445260A5073756389 @default.
- W3095445260 hasConcept C111335779 @default.
- W3095445260 hasConcept C112680207 @default.
- W3095445260 hasConcept C11413529 @default.
- W3095445260 hasConcept C114614502 @default.
- W3095445260 hasConcept C126255220 @default.
- W3095445260 hasConcept C149635348 @default.
- W3095445260 hasConcept C157972887 @default.
- W3095445260 hasConcept C184720557 @default.
- W3095445260 hasConcept C2524010 @default.
- W3095445260 hasConcept C26517878 @default.
- W3095445260 hasConcept C2780513914 @default.
- W3095445260 hasConcept C28855332 @default.
- W3095445260 hasConcept C33923547 @default.
- W3095445260 hasConcept C38652104 @default.
- W3095445260 hasConcept C41008148 @default.
- W3095445260 hasConceptScore W3095445260C111335779 @default.
- W3095445260 hasConceptScore W3095445260C112680207 @default.
- W3095445260 hasConceptScore W3095445260C11413529 @default.