Matches in SemOpenAlex for { <https://semopenalex.org/work/W3208265512> ?p ?o ?g. }
- W3208265512 abstract "Due to the explosion in the size of the training datasets, distributed learning has received growing interest in recent years. One of the major bottlenecks is the large communication cost between the central server and the local workers. While error feedback compression has been proven to be successful in reducing communication costs with stochastic gradient descent (SGD), there are much fewer attempts in building communication-efficient adaptive gradient methods with provable guarantees, which are widely used in training large-scale machine learning models. In this paper, we propose a new communication-compressed AMSGrad for distributed nonconvex optimization problem, which is provably efficient. Our proposed distributed learning framework features an effective gradient compression strategy and a worker-side model update design. We prove that the proposed communication-efficient distributed adaptive gradient method converges to the first-order stationary point with the same iteration complexity as uncompressed vanilla AMSGrad in the stochastic nonconvex optimization setting. Experiments on various benchmarks back up our theory." @default.
- W3208265512 created "2021-11-08" @default.
- W3208265512 creator A5006335513 @default.
- W3208265512 creator A5075760125 @default.
- W3208265512 creator A5082652766 @default.
- W3208265512 date "2021-11-01" @default.
- W3208265512 modified "2023-09-27" @default.
- W3208265512 title "Communication-Compressed Adaptive Gradient Method for Distributed Nonconvex Optimization." @default.
- W3208265512 cites W1522301498 @default.
- W3208265512 cites W1686810756 @default.
- W3208265512 cites W1994616650 @default.
- W3208265512 cites W2146502635 @default.
- W3208265512 cites W2153635508 @default.
- W3208265512 cites W2184045248 @default.
- W3208265512 cites W2194775991 @default.
- W3208265512 cites W2401231614 @default.
- W3208265512 cites W2407022425 @default.
- W3208265512 cites W2886463271 @default.
- W3208265512 cites W2890924858 @default.
- W3208265512 cites W2891952073 @default.
- W3208265512 cites W2908510526 @default.
- W3208265512 cites W2911863041 @default.
- W3208265512 cites W2931501180 @default.
- W3208265512 cites W2946815927 @default.
- W3208265512 cites W2949935872 @default.
- W3208265512 cites W2951651062 @default.
- W3208265512 cites W2952545433 @default.
- W3208265512 cites W2954365017 @default.
- W3208265512 cites W2963264832 @default.
- W3208265512 cites W2963341956 @default.
- W3208265512 cites W2963563140 @default.
- W3208265512 cites W2963664311 @default.
- W3208265512 cites W2971064744 @default.
- W3208265512 cites W2973096819 @default.
- W3208265512 cites W3007528421 @default.
- W3208265512 cites W3022625932 @default.
- W3208265512 cites W3030163527 @default.
- W3208265512 cites W3034995656 @default.
- W3208265512 cites W3035358246 @default.
- W3208265512 cites W3118608800 @default.
- W3208265512 cites W3127394918 @default.
- W3208265512 cites W3157279585 @default.
- W3208265512 cites W3166972615 @default.
- W3208265512 cites W6908809 @default.
- W3208265512 hasPublicationYear "2021" @default.
- W3208265512 type Work @default.
- W3208265512 sameAs 3208265512 @default.
- W3208265512 citedByCount "0" @default.
- W3208265512 crossrefType "posted-content" @default.
- W3208265512 hasAuthorship W3208265512A5006335513 @default.
- W3208265512 hasAuthorship W3208265512A5075760125 @default.
- W3208265512 hasAuthorship W3208265512A5082652766 @default.
- W3208265512 hasConcept C111919701 @default.
- W3208265512 hasConcept C120314980 @default.
- W3208265512 hasConcept C126255220 @default.
- W3208265512 hasConcept C149672232 @default.
- W3208265512 hasConcept C153258448 @default.
- W3208265512 hasConcept C154945302 @default.
- W3208265512 hasConcept C15744967 @default.
- W3208265512 hasConcept C159985019 @default.
- W3208265512 hasConcept C180016635 @default.
- W3208265512 hasConcept C192562407 @default.
- W3208265512 hasConcept C19417346 @default.
- W3208265512 hasConcept C206688291 @default.
- W3208265512 hasConcept C2779582901 @default.
- W3208265512 hasConcept C33923547 @default.
- W3208265512 hasConcept C41008148 @default.
- W3208265512 hasConcept C50644808 @default.
- W3208265512 hasConceptScore W3208265512C111919701 @default.
- W3208265512 hasConceptScore W3208265512C120314980 @default.
- W3208265512 hasConceptScore W3208265512C126255220 @default.
- W3208265512 hasConceptScore W3208265512C149672232 @default.
- W3208265512 hasConceptScore W3208265512C153258448 @default.
- W3208265512 hasConceptScore W3208265512C154945302 @default.
- W3208265512 hasConceptScore W3208265512C15744967 @default.
- W3208265512 hasConceptScore W3208265512C159985019 @default.
- W3208265512 hasConceptScore W3208265512C180016635 @default.
- W3208265512 hasConceptScore W3208265512C192562407 @default.
- W3208265512 hasConceptScore W3208265512C19417346 @default.
- W3208265512 hasConceptScore W3208265512C206688291 @default.
- W3208265512 hasConceptScore W3208265512C2779582901 @default.
- W3208265512 hasConceptScore W3208265512C33923547 @default.
- W3208265512 hasConceptScore W3208265512C41008148 @default.
- W3208265512 hasConceptScore W3208265512C50644808 @default.
- W3208265512 hasLocation W32082655121 @default.
- W3208265512 hasOpenAccess W3208265512 @default.
- W3208265512 hasPrimaryLocation W32082655121 @default.
- W3208265512 hasRelatedWork W1977179619 @default.
- W3208265512 hasRelatedWork W2008911925 @default.
- W3208265512 hasRelatedWork W2060979176 @default.
- W3208265512 hasRelatedWork W2079797353 @default.
- W3208265512 hasRelatedWork W2187389844 @default.
- W3208265512 hasRelatedWork W2288956432 @default.
- W3208265512 hasRelatedWork W2727721082 @default.
- W3208265512 hasRelatedWork W2743121208 @default.
- W3208265512 hasRelatedWork W2766258127 @default.
- W3208265512 hasRelatedWork W2783162894 @default.
- W3208265512 hasRelatedWork W2794270121 @default.
- W3208265512 hasRelatedWork W2799626538 @default.
- W3208265512 hasRelatedWork W2964258970 @default.