Matches in SemOpenAlex for { <https://semopenalex.org/work/W3203246084> ?p ?o ?g. }
Showing items 1 to 85 of
85
with 100 items per page.
- W3203246084 endingPage "11" @default.
- W3203246084 startingPage "1" @default.
- W3203246084 abstract "Graph-level representation learning is the pivotal step for downstream tasks that operate on the whole graph. The most common approach to this problem is graph pooling, where node features are typically averaged or summed to obtain the graph representations. However, pooling operations like averaging or summing inevitably cause severe information missing, which may severely downgrade the final performance. In this paper, we argue what is crucial to graph-level downstream tasks includes not only the topological structure but also the <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>distribution</i> from which nodes are sampled. Therefore, powered by existing Graph Neural Networks (GNN), we propose a new plug-and-play pooling module, termed as <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>Distribution Knowledge Embedding</i> (DKEPool), where graphs are viewed as distributions on top of GNNs and the pooling goal is to summarize the entire distribution information instead of retaining a certain feature vector by simple predefined pooling operations. A DKEPool network <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>de facto</i> disassembles representation learning into two stages, <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>structure learning</i> and <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>distribution learning</i> . Structure learning follows a recursive neighborhood aggregation scheme to update node features where structure information is obtained. Distribution learning, on the other hand, omits node interconnections and focuses more on the distribution depicted by all the nodes. Extensive experiments on graph classification tasks demonstrate that the proposed DKEPool significantly and consistently outperforms the state-of-the-art methods. The code is avaliable at <uri xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>https://github.com/chenchkx/dkepool</uri>" @default.
- W3203246084 created "2021-10-11" @default.
- W3203246084 creator A5006824735 @default.
- W3203246084 creator A5010740000 @default.
- W3203246084 creator A5016449976 @default.
- W3203246084 creator A5026532752 @default.
- W3203246084 creator A5043914348 @default.
- W3203246084 creator A5047371218 @default.
- W3203246084 creator A5090854835 @default.
- W3203246084 date "2022-01-01" @default.
- W3203246084 modified "2023-09-26" @default.
- W3203246084 title "Distribution Knowledge Embedding for Graph Pooling" @default.
- W3203246084 cites W1996955012 @default.
- W3203246084 cites W2006766164 @default.
- W3203246084 cites W2008857988 @default.
- W3203246084 cites W2022416612 @default.
- W3203246084 cites W2037900003 @default.
- W3203246084 cites W2343076010 @default.
- W3203246084 cites W2423822072 @default.
- W3203246084 cites W2601789736 @default.
- W3203246084 cites W2747623286 @default.
- W3203246084 cites W2749507674 @default.
- W3203246084 cites W2752965000 @default.
- W3203246084 cites W2904218366 @default.
- W3203246084 cites W2962761264 @default.
- W3203246084 cites W2964009357 @default.
- W3203246084 cites W2997997679 @default.
- W3203246084 cites W2998008360 @default.
- W3203246084 cites W2999682157 @default.
- W3203246084 cites W3027380106 @default.
- W3203246084 cites W3034090344 @default.
- W3203246084 cites W3113037084 @default.
- W3203246084 cites W3133780103 @default.
- W3203246084 cites W3143205592 @default.
- W3203246084 cites W3156823866 @default.
- W3203246084 cites W3208074551 @default.
- W3203246084 doi "https://doi.org/10.1109/tkde.2022.3208063" @default.
- W3203246084 hasPublicationYear "2022" @default.
- W3203246084 type Work @default.
- W3203246084 sameAs 3203246084 @default.
- W3203246084 citedByCount "1" @default.
- W3203246084 countsByYear W32032460842023 @default.
- W3203246084 crossrefType "journal-article" @default.
- W3203246084 hasAuthorship W3203246084A5006824735 @default.
- W3203246084 hasAuthorship W3203246084A5010740000 @default.
- W3203246084 hasAuthorship W3203246084A5016449976 @default.
- W3203246084 hasAuthorship W3203246084A5026532752 @default.
- W3203246084 hasAuthorship W3203246084A5043914348 @default.
- W3203246084 hasAuthorship W3203246084A5047371218 @default.
- W3203246084 hasAuthorship W3203246084A5090854835 @default.
- W3203246084 hasBestOaLocation W32032460842 @default.
- W3203246084 hasConcept C119857082 @default.
- W3203246084 hasConcept C132525143 @default.
- W3203246084 hasConcept C154945302 @default.
- W3203246084 hasConcept C41008148 @default.
- W3203246084 hasConcept C41608201 @default.
- W3203246084 hasConcept C70437156 @default.
- W3203246084 hasConcept C80444323 @default.
- W3203246084 hasConceptScore W3203246084C119857082 @default.
- W3203246084 hasConceptScore W3203246084C132525143 @default.
- W3203246084 hasConceptScore W3203246084C154945302 @default.
- W3203246084 hasConceptScore W3203246084C41008148 @default.
- W3203246084 hasConceptScore W3203246084C41608201 @default.
- W3203246084 hasConceptScore W3203246084C70437156 @default.
- W3203246084 hasConceptScore W3203246084C80444323 @default.
- W3203246084 hasLocation W32032460841 @default.
- W3203246084 hasLocation W32032460842 @default.
- W3203246084 hasOpenAccess W3203246084 @default.
- W3203246084 hasPrimaryLocation W32032460841 @default.
- W3203246084 hasRelatedWork W2923818335 @default.
- W3203246084 hasRelatedWork W3035116611 @default.
- W3203246084 hasRelatedWork W3044354590 @default.
- W3203246084 hasRelatedWork W3093454656 @default.
- W3203246084 hasRelatedWork W3094605108 @default.
- W3203246084 hasRelatedWork W3145941582 @default.
- W3203246084 hasRelatedWork W4226361842 @default.
- W3203246084 hasRelatedWork W4287763734 @default.
- W3203246084 hasRelatedWork W4296821948 @default.
- W3203246084 hasRelatedWork W4310879833 @default.
- W3203246084 isParatext "false" @default.
- W3203246084 isRetracted "false" @default.
- W3203246084 magId "3203246084" @default.
- W3203246084 workType "article" @default.