Matches in SemOpenAlex for { <https://semopenalex.org/work/W4289598986> ?p ?o ?g. }
Showing items 1 to 80 of
80
with 100 items per page.
- W4289598986 abstract "This work introduces DADAO: the first decentralized, accelerated, asynchronous, primal, first-order algorithm to minimize a sum of $L$-smooth and $mu$-strongly convex functions distributed over a given network of size $n$. Our key insight is based on modeling the local gradient updates and gossip communication procedures with separate independent Poisson Point Processes. This allows us to decouple the computation and communication steps, which can be run in parallel, while making the whole approach completely asynchronous, leading to communication acceleration compared to synchronous approaches. Our new method employs primal gradients and does not use a multi-consensus inner loop nor other ad-hoc mechanisms such as Error Feedback, Gradient Tracking, or a Proximal operator. By relating the inverse of the smallest positive eigenvalue of the Laplacian matrix $chi_1$ and the maximal resistance $chi_2leq chi_1$ of the graph to a sufficient minimal communication rate between the nodes of the network, we show that our algorithm requires $mathcal{O}(nsqrt{frac{L}{mu}}log(frac{1}{epsilon}))$ local gradients and only $mathcal{O}(nsqrt{chi_1chi_2}sqrt{frac{L}{mu}}log(frac{1}{epsilon}))$ communications to reach a precision $epsilon$, up to logarithmic terms. Thus, we simultaneously obtain an accelerated rate for both computations and communications, leading to an improvement over state-of-the-art works, our simulations further validating the strength of our relatively unconstrained method. We also propose a SDP relaxation to find the optimal gossip rate of each edge minimizing the total number of communications for a given graph, resulting in faster convergence compared to standard approaches relying on uniform communication weights. Our source code is released on a public repository." @default.
- W4289598986 created "2022-08-03" @default.
- W4289598986 creator A5045828346 @default.
- W4289598986 creator A5086553821 @default.
- W4289598986 date "2022-07-26" @default.
- W4289598986 modified "2023-09-24" @default.
- W4289598986 title "DADAO: Decoupled Accelerated Decentralized Asynchronous Optimization" @default.
- W4289598986 doi "https://doi.org/10.48550/arxiv.2208.00779" @default.
- W4289598986 hasPublicationYear "2022" @default.
- W4289598986 type Work @default.
- W4289598986 citedByCount "0" @default.
- W4289598986 crossrefType "posted-content" @default.
- W4289598986 hasAuthorship W4289598986A5045828346 @default.
- W4289598986 hasAuthorship W4289598986A5086553821 @default.
- W4289598986 hasBestOaLocation W42895989861 @default.
- W4289598986 hasConcept C112680207 @default.
- W4289598986 hasConcept C11413529 @default.
- W4289598986 hasConcept C114614502 @default.
- W4289598986 hasConcept C115178988 @default.
- W4289598986 hasConcept C118615104 @default.
- W4289598986 hasConcept C121332964 @default.
- W4289598986 hasConcept C130120984 @default.
- W4289598986 hasConcept C132525143 @default.
- W4289598986 hasConcept C134306372 @default.
- W4289598986 hasConcept C145446738 @default.
- W4289598986 hasConcept C151319957 @default.
- W4289598986 hasConcept C158693339 @default.
- W4289598986 hasConcept C165700671 @default.
- W4289598986 hasConcept C184720557 @default.
- W4289598986 hasConcept C199360897 @default.
- W4289598986 hasConcept C207467116 @default.
- W4289598986 hasConcept C2524010 @default.
- W4289598986 hasConcept C2777818330 @default.
- W4289598986 hasConcept C33923547 @default.
- W4289598986 hasConcept C39927690 @default.
- W4289598986 hasConcept C41008148 @default.
- W4289598986 hasConcept C45374587 @default.
- W4289598986 hasConcept C62520636 @default.
- W4289598986 hasConcept C76155785 @default.
- W4289598986 hasConceptScore W4289598986C112680207 @default.
- W4289598986 hasConceptScore W4289598986C11413529 @default.
- W4289598986 hasConceptScore W4289598986C114614502 @default.
- W4289598986 hasConceptScore W4289598986C115178988 @default.
- W4289598986 hasConceptScore W4289598986C118615104 @default.
- W4289598986 hasConceptScore W4289598986C121332964 @default.
- W4289598986 hasConceptScore W4289598986C130120984 @default.
- W4289598986 hasConceptScore W4289598986C132525143 @default.
- W4289598986 hasConceptScore W4289598986C134306372 @default.
- W4289598986 hasConceptScore W4289598986C145446738 @default.
- W4289598986 hasConceptScore W4289598986C151319957 @default.
- W4289598986 hasConceptScore W4289598986C158693339 @default.
- W4289598986 hasConceptScore W4289598986C165700671 @default.
- W4289598986 hasConceptScore W4289598986C184720557 @default.
- W4289598986 hasConceptScore W4289598986C199360897 @default.
- W4289598986 hasConceptScore W4289598986C207467116 @default.
- W4289598986 hasConceptScore W4289598986C2524010 @default.
- W4289598986 hasConceptScore W4289598986C2777818330 @default.
- W4289598986 hasConceptScore W4289598986C33923547 @default.
- W4289598986 hasConceptScore W4289598986C39927690 @default.
- W4289598986 hasConceptScore W4289598986C41008148 @default.
- W4289598986 hasConceptScore W4289598986C45374587 @default.
- W4289598986 hasConceptScore W4289598986C62520636 @default.
- W4289598986 hasConceptScore W4289598986C76155785 @default.
- W4289598986 hasLocation W42895989861 @default.
- W4289598986 hasLocation W42895989862 @default.
- W4289598986 hasOpenAccess W4289598986 @default.
- W4289598986 hasPrimaryLocation W42895989861 @default.
- W4289598986 hasRelatedWork W1832592836 @default.
- W4289598986 hasRelatedWork W2096663817 @default.
- W4289598986 hasRelatedWork W2298820818 @default.
- W4289598986 hasRelatedWork W2343714073 @default.
- W4289598986 hasRelatedWork W2732469621 @default.
- W4289598986 hasRelatedWork W2786674967 @default.
- W4289598986 hasRelatedWork W2951938077 @default.
- W4289598986 hasRelatedWork W3132233781 @default.
- W4289598986 hasRelatedWork W4293876796 @default.
- W4289598986 hasRelatedWork W4300525916 @default.
- W4289598986 isParatext "false" @default.
- W4289598986 isRetracted "false" @default.
- W4289598986 workType "article" @default.