FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3200726313> ?p ?o ?g. }

• W3200726313 abstract "Graph Neural Networks (GNNs) extend basic Neural Networks (NNs) by using the graph structures based on the relational inductive bias (homophily assumption). Though GNNs are believed to outperform NNs in real-world tasks, performance advantages of GNNs over graph-agnostic NNs seem not generally satisfactory. Heterophily has been considered as a main cause and numerous works have been put forward to address it. In this paper, we first show that not all cases of heterophily are harmful for GNNs with aggregation operation. Then, we propose new metrics based on a similarity matrix which considers the influence of both graph structure and input features on GNNs. The metrics demonstrate advantages over the commonly used homophily metrics by tests on synthetic graphs. From the metrics and the observations, we find some cases of harmful heterophily can be addressed by diversification operation. With this fact and knowledge of filterbanks, we propose the Adaptive Channel Mixing (ACM) framework to adaptively exploit aggregation, diversification and identity channels in each GNN layer to address harmful heterophily. We validate the ACM-augmented baselines with 10 real-world node classification tasks. They consistently achieve significant performance gain and exceed the state-of-the-art GNNs on most of the tasks without incurring significant computational burden." @default.
• W3200726313 created "2021-09-27" @default.
• W3200726313 creator A5008039784 @default.
• W3200726313 creator A5009954522 @default.
• W3200726313 creator A5017824411 @default.
• W3200726313 creator A5043314002 @default.
• W3200726313 creator A5048761975 @default.
• W3200726313 creator A5062087251 @default.
• W3200726313 creator A5065836447 @default.
• W3200726313 creator A5083343995 @default.
• W3200726313 date "2021-09-12" @default.
• W3200726313 modified "2023-09-27" @default.
• W3200726313 title "Is Heterophily A Real Nightmare For Graph Neural Networks To Do Node Classification?" @default.
• W3200726313 cites W1578099820 @default.
• W3200726313 cites W2112796928 @default.
• W3200726313 cites W2116341502 @default.
• W3200726313 cites W2130354913 @default.
• W3200726313 cites W2143612262 @default.
• W3200726313 cites W2163605009 @default.
• W3200726313 cites W2220894258 @default.
• W3200726313 cites W2519887557 @default.
• W3200726313 cites W2558748708 @default.
• W3200726313 cites W2624431344 @default.
• W3200726313 cites W2805516822 @default.
• W3200726313 cites W2908510526 @default.
• W3200726313 cites W2916106175 @default.
• W3200726313 cites W2918342466 @default.
• W3200726313 cites W2919115771 @default.
• W3200726313 cites W2945848398 @default.
• W3200726313 cites W2949406091 @default.
• W3200726313 cites W2949945331 @default.
• W3200726313 cites W2963858333 @default.
• W3200726313 cites W2963893572 @default.
• W3200726313 cites W2964051675 @default.
• W3200726313 cites W2964114465 @default.
• W3200726313 cites W2964121744 @default.
• W3200726313 cites W2964308564 @default.
• W3200726313 cites W2964321699 @default.
• W3200726313 cites W2971328361 @default.
• W3200726313 cites W2996084050 @default.
• W3200726313 cites W3005644236 @default.
• W3200726313 cites W3030661718 @default.
• W3200726313 cites W3034492151 @default.
• W3200726313 cites W3051256797 @default.
• W3200726313 cites W3087257704 @default.
• W3200726313 cites W3090999459 @default.
• W3200726313 cites W3100646853 @default.
• W3200726313 cites W3103557031 @default.
• W3200726313 cites W3120233338 @default.
• W3200726313 cites W3133222630 @default.
• W3200726313 cites W3144257051 @default.
• W3200726313 cites W3173240144 @default.
• W3200726313 cites W3177075738 @default.
• W3200726313 doi "https://doi.org/10.48550/arxiv.2109.05641" @default.
• W3200726313 hasPublicationYear "2021" @default.
• W3200726313 type Work @default.
• W3200726313 sameAs 3200726313 @default.
• W3200726313 citedByCount "1" @default.
• W3200726313 countsByYear W32007263132021 @default.
• W3200726313 crossrefType "posted-content" @default.
• W3200726313 hasAuthorship W3200726313A5008039784 @default.
• W3200726313 hasAuthorship W3200726313A5009954522 @default.
• W3200726313 hasAuthorship W3200726313A5017824411 @default.
• W3200726313 hasAuthorship W3200726313A5043314002 @default.
• W3200726313 hasAuthorship W3200726313A5048761975 @default.
• W3200726313 hasAuthorship W3200726313A5062087251 @default.
• W3200726313 hasAuthorship W3200726313A5065836447 @default.
• W3200726313 hasAuthorship W3200726313A5083343995 @default.
• W3200726313 hasBestOaLocation W32007263131 @default.
• W3200726313 hasConcept C114614502 @default.
• W3200726313 hasConcept C119857082 @default.
• W3200726313 hasConcept C132525143 @default.
• W3200726313 hasConcept C154945302 @default.
• W3200726313 hasConcept C165696696 @default.
• W3200726313 hasConcept C2779812341 @default.
• W3200726313 hasConcept C33923547 @default.
• W3200726313 hasConcept C38652104 @default.
• W3200726313 hasConcept C41008148 @default.
• W3200726313 hasConcept C50644808 @default.
• W3200726313 hasConcept C80444323 @default.
• W3200726313 hasConceptScore W3200726313C114614502 @default.
• W3200726313 hasConceptScore W3200726313C119857082 @default.
• W3200726313 hasConceptScore W3200726313C132525143 @default.
• W3200726313 hasConceptScore W3200726313C154945302 @default.
• W3200726313 hasConceptScore W3200726313C165696696 @default.
• W3200726313 hasConceptScore W3200726313C2779812341 @default.
• W3200726313 hasConceptScore W3200726313C33923547 @default.
• W3200726313 hasConceptScore W3200726313C38652104 @default.
• W3200726313 hasConceptScore W3200726313C41008148 @default.
• W3200726313 hasConceptScore W3200726313C50644808 @default.
• W3200726313 hasConceptScore W3200726313C80444323 @default.
• W3200726313 hasLocation W32007263131 @default.
• W3200726313 hasOpenAccess W3200726313 @default.
• W3200726313 hasPrimaryLocation W32007263131 @default.
• W3200726313 hasRelatedWork W1507793640 @default.
• W3200726313 hasRelatedWork W2952419077 @default.
• W3200726313 hasRelatedWork W4200632081 @default.
• W3200726313 hasRelatedWork W4226197742 @default.
• W3200726313 hasRelatedWork W4297899248 @default.
• W3200726313 hasRelatedWork W4306920765 @default.

• next