Matches in SemOpenAlex for { <https://semopenalex.org/work/W4366733367> ?p ?o ?g. }
Showing items 1 to 73 of
73
with 100 items per page.
- W4366733367 abstract "Graph neural networks (GNN) suffer from severe inefficiency. It is mainly caused by the exponential growth of node dependency with the increase of layers. It extremely limits the application of stochastic optimization algorithms so that the training of GNN is usually time-consuming. To address this problem, we propose to decouple a multi-layer GNN as multiple simple modules for more efficient training, which is comprised of classical forward training (FT)and designed backward training (BT). Under the proposed framework, each module can be trained efficiently in FT by stochastic algorithms without distortion of graph information owing to its simplicity. To avoid the only unidirectional information delivery of FT and sufficiently train shallow modules with the deeper ones, we develop a backward training mechanism that makes the former modules perceive the latter modules. The backward training introduces the reversed information delivery into the decoupled modules as well as the forward information delivery. To investigate how the decoupling and greedy training affect the representational capacity, we theoretically prove that the error produced by linear modules will not accumulate on unsupervised tasks in most cases. The theoretical and experimental results show that the proposed framework is highly efficient with reasonable performance." @default.
- W4366733367 created "2023-04-24" @default.
- W4366733367 creator A5004015448 @default.
- W4366733367 creator A5029601834 @default.
- W4366733367 creator A5068918243 @default.
- W4366733367 date "2023-04-20" @default.
- W4366733367 modified "2023-09-26" @default.
- W4366733367 title "Decouple Graph Neural Networks: Train Multiple Simple GNNs Simultaneously Instead of One" @default.
- W4366733367 doi "https://doi.org/10.48550/arxiv.2304.10126" @default.
- W4366733367 hasPublicationYear "2023" @default.
- W4366733367 type Work @default.
- W4366733367 citedByCount "0" @default.
- W4366733367 crossrefType "posted-content" @default.
- W4366733367 hasAuthorship W4366733367A5004015448 @default.
- W4366733367 hasAuthorship W4366733367A5029601834 @default.
- W4366733367 hasAuthorship W4366733367A5068918243 @default.
- W4366733367 hasBestOaLocation W43667333671 @default.
- W4366733367 hasConcept C111472728 @default.
- W4366733367 hasConcept C11413529 @default.
- W4366733367 hasConcept C120314980 @default.
- W4366733367 hasConcept C126255220 @default.
- W4366733367 hasConcept C127413603 @default.
- W4366733367 hasConcept C132525143 @default.
- W4366733367 hasConcept C133731056 @default.
- W4366733367 hasConcept C138885662 @default.
- W4366733367 hasConcept C154945302 @default.
- W4366733367 hasConcept C162324750 @default.
- W4366733367 hasConcept C16311509 @default.
- W4366733367 hasConcept C175444787 @default.
- W4366733367 hasConcept C19768560 @default.
- W4366733367 hasConcept C205606062 @default.
- W4366733367 hasConcept C2778869765 @default.
- W4366733367 hasConcept C2780586882 @default.
- W4366733367 hasConcept C33923547 @default.
- W4366733367 hasConcept C41008148 @default.
- W4366733367 hasConcept C50644808 @default.
- W4366733367 hasConcept C80444323 @default.
- W4366733367 hasConceptScore W4366733367C111472728 @default.
- W4366733367 hasConceptScore W4366733367C11413529 @default.
- W4366733367 hasConceptScore W4366733367C120314980 @default.
- W4366733367 hasConceptScore W4366733367C126255220 @default.
- W4366733367 hasConceptScore W4366733367C127413603 @default.
- W4366733367 hasConceptScore W4366733367C132525143 @default.
- W4366733367 hasConceptScore W4366733367C133731056 @default.
- W4366733367 hasConceptScore W4366733367C138885662 @default.
- W4366733367 hasConceptScore W4366733367C154945302 @default.
- W4366733367 hasConceptScore W4366733367C162324750 @default.
- W4366733367 hasConceptScore W4366733367C16311509 @default.
- W4366733367 hasConceptScore W4366733367C175444787 @default.
- W4366733367 hasConceptScore W4366733367C19768560 @default.
- W4366733367 hasConceptScore W4366733367C205606062 @default.
- W4366733367 hasConceptScore W4366733367C2778869765 @default.
- W4366733367 hasConceptScore W4366733367C2780586882 @default.
- W4366733367 hasConceptScore W4366733367C33923547 @default.
- W4366733367 hasConceptScore W4366733367C41008148 @default.
- W4366733367 hasConceptScore W4366733367C50644808 @default.
- W4366733367 hasConceptScore W4366733367C80444323 @default.
- W4366733367 hasLocation W43667333671 @default.
- W4366733367 hasOpenAccess W4366733367 @default.
- W4366733367 hasPrimaryLocation W43667333671 @default.
- W4366733367 hasRelatedWork W2131662294 @default.
- W4366733367 hasRelatedWork W2369797701 @default.
- W4366733367 hasRelatedWork W2379773790 @default.
- W4366733367 hasRelatedWork W2381168281 @default.
- W4366733367 hasRelatedWork W2430210575 @default.
- W4366733367 hasRelatedWork W2612062150 @default.
- W4366733367 hasRelatedWork W3013050996 @default.
- W4366733367 hasRelatedWork W3200266554 @default.
- W4366733367 hasRelatedWork W3203251273 @default.
- W4366733367 hasRelatedWork W3210155192 @default.
- W4366733367 isParatext "false" @default.
- W4366733367 isRetracted "false" @default.
- W4366733367 workType "article" @default.