Matches in SemOpenAlex for { <https://semopenalex.org/work/W4318189513> ?p ?o ?g. }
Showing items 1 to 86 of
86
with 100 items per page.
- W4318189513 endingPage "201" @default.
- W4318189513 startingPage "189" @default.
- W4318189513 abstract "Noisy gradient algorithms have emerged as one of the most popular algorithms for distributed optimization with massive data. Choosing proper step-size schedules is an important task to tune in the algorithms for good performance. For the algorithms to attain fast convergence and high accuracy, it is intuitive to use large step-sizes in the initial iterations when the gradient noise is typically small compared to the algorithm-steps, and reduce the step-sizes as the algorithm progresses. This intuition has been confirmed in theory and practice for stochastic gradient descent. However, similar results are lacking for other methods using approximate gradients. This paper shows that the diminishing step-size strategies can indeed be applied for a broad class of noisy gradient algorithms. Our analysis framework is based on two classes of systems that characterize the impact of the step-sizes on the convergence performance of many algorithms. Our results show that such step-size schedules enable these algorithms to enjoy the optimal rate. We exemplify our results on stochastic compression algorithms. Our experiments validate fast convergence of these algorithms with the step decay schedules." @default.
- W4318189513 created "2023-01-27" @default.
- W4318189513 creator A5024256519 @default.
- W4318189513 creator A5024792434 @default.
- W4318189513 creator A5027182415 @default.
- W4318189513 creator A5054704457 @default.
- W4318189513 date "2023-01-01" @default.
- W4318189513 modified "2023-10-03" @default.
- W4318189513 title "Improved Step-Size Schedules for Proximal Noisy Gradient Methods" @default.
- W4318189513 cites W1992208280 @default.
- W4318189513 cites W1994616650 @default.
- W4318189513 cites W2149479912 @default.
- W4318189513 cites W2153635508 @default.
- W4318189513 cites W2543927648 @default.
- W4318189513 cites W2746600820 @default.
- W4318189513 cites W2748196763 @default.
- W4318189513 cites W2963190258 @default.
- W4318189513 cites W2963307318 @default.
- W4318189513 cites W2963470657 @default.
- W4318189513 cites W2980398138 @default.
- W4318189513 cites W3013797917 @default.
- W4318189513 cites W3023838981 @default.
- W4318189513 cites W3162168477 @default.
- W4318189513 cites W3176526097 @default.
- W4318189513 cites W3198503994 @default.
- W4318189513 cites W4248800565 @default.
- W4318189513 cites W4320147968 @default.
- W4318189513 doi "https://doi.org/10.1109/tsp.2023.3237392" @default.
- W4318189513 hasPublicationYear "2023" @default.
- W4318189513 type Work @default.
- W4318189513 citedByCount "0" @default.
- W4318189513 crossrefType "journal-article" @default.
- W4318189513 hasAuthorship W4318189513A5024256519 @default.
- W4318189513 hasAuthorship W4318189513A5024792434 @default.
- W4318189513 hasAuthorship W4318189513A5027182415 @default.
- W4318189513 hasAuthorship W4318189513A5054704457 @default.
- W4318189513 hasConcept C10494615 @default.
- W4318189513 hasConcept C11413529 @default.
- W4318189513 hasConcept C126255220 @default.
- W4318189513 hasConcept C127162648 @default.
- W4318189513 hasConcept C153258448 @default.
- W4318189513 hasConcept C154945302 @default.
- W4318189513 hasConcept C162324750 @default.
- W4318189513 hasConcept C206688291 @default.
- W4318189513 hasConcept C2777303404 @default.
- W4318189513 hasConcept C31258907 @default.
- W4318189513 hasConcept C33923547 @default.
- W4318189513 hasConcept C41008148 @default.
- W4318189513 hasConcept C50522688 @default.
- W4318189513 hasConcept C50644808 @default.
- W4318189513 hasConcept C57869625 @default.
- W4318189513 hasConceptScore W4318189513C10494615 @default.
- W4318189513 hasConceptScore W4318189513C11413529 @default.
- W4318189513 hasConceptScore W4318189513C126255220 @default.
- W4318189513 hasConceptScore W4318189513C127162648 @default.
- W4318189513 hasConceptScore W4318189513C153258448 @default.
- W4318189513 hasConceptScore W4318189513C154945302 @default.
- W4318189513 hasConceptScore W4318189513C162324750 @default.
- W4318189513 hasConceptScore W4318189513C206688291 @default.
- W4318189513 hasConceptScore W4318189513C2777303404 @default.
- W4318189513 hasConceptScore W4318189513C31258907 @default.
- W4318189513 hasConceptScore W4318189513C33923547 @default.
- W4318189513 hasConceptScore W4318189513C41008148 @default.
- W4318189513 hasConceptScore W4318189513C50522688 @default.
- W4318189513 hasConceptScore W4318189513C50644808 @default.
- W4318189513 hasConceptScore W4318189513C57869625 @default.
- W4318189513 hasFunder F4320322581 @default.
- W4318189513 hasLocation W43181895131 @default.
- W4318189513 hasOpenAccess W4318189513 @default.
- W4318189513 hasPrimaryLocation W43181895131 @default.
- W4318189513 hasRelatedWork W2075181955 @default.
- W4318189513 hasRelatedWork W2128886773 @default.
- W4318189513 hasRelatedWork W2170262790 @default.
- W4318189513 hasRelatedWork W2766590049 @default.
- W4318189513 hasRelatedWork W2766636532 @default.
- W4318189513 hasRelatedWork W2928207644 @default.
- W4318189513 hasRelatedWork W2995653155 @default.
- W4318189513 hasRelatedWork W3114199884 @default.
- W4318189513 hasRelatedWork W4221139735 @default.
- W4318189513 hasRelatedWork W4307415224 @default.
- W4318189513 hasVolume "71" @default.
- W4318189513 isParatext "false" @default.
- W4318189513 isRetracted "false" @default.
- W4318189513 workType "article" @default.