Matches in SemOpenAlex for { <https://semopenalex.org/work/W4361195852> ?p ?o ?g. }
Showing items 1 to 79 of
79
with 100 items per page.
- W4361195852 abstract "Large-scale graphs with billions of edges are ubiquitous in many industries, science, and engineering fields such as recommendation systems, social graph analysis, knowledge base, material science, and biology. Graph neural networks (GNN), an emerging class of machine learning models, are increasingly adopted to learn on these graphs due to their superior performance in various graph analytics tasks. Mini-batch training is commonly adopted to train on large graphs, and data parallelism is the standard approach to scale mini-batch training to multiple GPUs. In this paper, we argue that several fundamental performance bottlenecks of GNN training systems have to do with inherent limitations of the data parallel approach. We then propose split parallelism, a novel parallel mini-batch training paradigm. We implement split parallelism in a novel system called gsplit and show that it outperforms state-of-the-art systems such as DGL, Quiver, and PaGraph." @default.
- W4361195852 created "2023-03-31" @default.
- W4361195852 creator A5005127609 @default.
- W4361195852 creator A5010813852 @default.
- W4361195852 creator A5019474580 @default.
- W4361195852 creator A5029408111 @default.
- W4361195852 creator A5029824262 @default.
- W4361195852 creator A5057065275 @default.
- W4361195852 creator A5068345732 @default.
- W4361195852 date "2023-03-23" @default.
- W4361195852 modified "2023-09-26" @default.
- W4361195852 title "GSplit: Scaling Graph Neural Network Training on Large Graphs via Split-Parallelism" @default.
- W4361195852 doi "https://doi.org/10.48550/arxiv.2303.13775" @default.
- W4361195852 hasPublicationYear "2023" @default.
- W4361195852 type Work @default.
- W4361195852 citedByCount "0" @default.
- W4361195852 crossrefType "posted-content" @default.
- W4361195852 hasAuthorship W4361195852A5005127609 @default.
- W4361195852 hasAuthorship W4361195852A5010813852 @default.
- W4361195852 hasAuthorship W4361195852A5019474580 @default.
- W4361195852 hasAuthorship W4361195852A5029408111 @default.
- W4361195852 hasAuthorship W4361195852A5029824262 @default.
- W4361195852 hasAuthorship W4361195852A5057065275 @default.
- W4361195852 hasAuthorship W4361195852A5068345732 @default.
- W4361195852 hasBestOaLocation W43611958521 @default.
- W4361195852 hasConcept C127413603 @default.
- W4361195852 hasConcept C132525143 @default.
- W4361195852 hasConcept C154945302 @default.
- W4361195852 hasConcept C173608175 @default.
- W4361195852 hasConcept C176225458 @default.
- W4361195852 hasConcept C2522767166 @default.
- W4361195852 hasConcept C2524010 @default.
- W4361195852 hasConcept C2781172179 @default.
- W4361195852 hasConcept C2993955422 @default.
- W4361195852 hasConcept C33923547 @default.
- W4361195852 hasConcept C41008148 @default.
- W4361195852 hasConcept C42992933 @default.
- W4361195852 hasConcept C50644808 @default.
- W4361195852 hasConcept C51632099 @default.
- W4361195852 hasConcept C55587333 @default.
- W4361195852 hasConcept C61483411 @default.
- W4361195852 hasConcept C79158427 @default.
- W4361195852 hasConcept C80444323 @default.
- W4361195852 hasConcept C99844830 @default.
- W4361195852 hasConceptScore W4361195852C127413603 @default.
- W4361195852 hasConceptScore W4361195852C132525143 @default.
- W4361195852 hasConceptScore W4361195852C154945302 @default.
- W4361195852 hasConceptScore W4361195852C173608175 @default.
- W4361195852 hasConceptScore W4361195852C176225458 @default.
- W4361195852 hasConceptScore W4361195852C2522767166 @default.
- W4361195852 hasConceptScore W4361195852C2524010 @default.
- W4361195852 hasConceptScore W4361195852C2781172179 @default.
- W4361195852 hasConceptScore W4361195852C2993955422 @default.
- W4361195852 hasConceptScore W4361195852C33923547 @default.
- W4361195852 hasConceptScore W4361195852C41008148 @default.
- W4361195852 hasConceptScore W4361195852C42992933 @default.
- W4361195852 hasConceptScore W4361195852C50644808 @default.
- W4361195852 hasConceptScore W4361195852C51632099 @default.
- W4361195852 hasConceptScore W4361195852C55587333 @default.
- W4361195852 hasConceptScore W4361195852C61483411 @default.
- W4361195852 hasConceptScore W4361195852C79158427 @default.
- W4361195852 hasConceptScore W4361195852C80444323 @default.
- W4361195852 hasConceptScore W4361195852C99844830 @default.
- W4361195852 hasLocation W43611958521 @default.
- W4361195852 hasOpenAccess W4361195852 @default.
- W4361195852 hasPrimaryLocation W43611958521 @default.
- W4361195852 hasRelatedWork W1496703677 @default.
- W4361195852 hasRelatedWork W1508832769 @default.
- W4361195852 hasRelatedWork W1595151633 @default.
- W4361195852 hasRelatedWork W1784521533 @default.
- W4361195852 hasRelatedWork W2047301304 @default.
- W4361195852 hasRelatedWork W2050855393 @default.
- W4361195852 hasRelatedWork W2378666660 @default.
- W4361195852 hasRelatedWork W3085352435 @default.
- W4361195852 hasRelatedWork W3095555187 @default.
- W4361195852 hasRelatedWork W4240606930 @default.
- W4361195852 isParatext "false" @default.
- W4361195852 isRetracted "false" @default.
- W4361195852 workType "article" @default.