Matches in SemOpenAlex for { <https://semopenalex.org/work/W3040986370> ?p ?o ?g. }
- W3040986370 abstract "Neural networks are often represented as graphs of connections between neurons. However, despite their wide use, there is currently little understanding of the relationship between the graph structure of the neural network and its predictive performance. Here we systematically investigate how does the graph structure of neural networks affect their predictive performance. To this end, we develop a novel graph-based representation of neural networks called relational graph, where layers of neural network computation correspond to rounds of message exchange along the graph structure. Using this representation we show that: (1) a sweet spot of relational graphs leads to neural networks with significantly improved predictive performance; (2) neural network's performance is approximately a smooth function of the clustering coefficient and average path length of its relational graph; (3) our findings are consistent across many different tasks and datasets; (4) the sweet spot can be identified efficiently; (5) top-performing neural networks have graph structure surprisingly similar to those of real biological neural networks. Our work opens new directions for the design of neural architectures and the understanding on neural networks in general." @default.
- W3040986370 created "2020-07-16" @default.
- W3040986370 creator A5003491365 @default.
- W3040986370 creator A5024785349 @default.
- W3040986370 creator A5049098969 @default.
- W3040986370 creator A5091272738 @default.
- W3040986370 date "2020-07-13" @default.
- W3040986370 modified "2023-09-25" @default.
- W3040986370 title "Graph Structure of Neural Networks" @default.
- W3040986370 cites W1529495223 @default.
- W3040986370 cites W2029949252 @default.
- W3040986370 cites W2112796928 @default.
- W3040986370 cites W2124637492 @default.
- W3040986370 cites W2125993726 @default.
- W3040986370 cites W2156387975 @default.
- W3040986370 cites W2163605009 @default.
- W3040986370 cites W2194775991 @default.
- W3040986370 cites W2518108298 @default.
- W3040986370 cites W2531409750 @default.
- W3040986370 cites W2549139847 @default.
- W3040986370 cites W2553303224 @default.
- W3040986370 cites W2612445135 @default.
- W3040986370 cites W2767544505 @default.
- W3040986370 cites W2785366763 @default.
- W3040986370 cites W2786103815 @default.
- W3040986370 cites W2788014331 @default.
- W3040986370 cites W2806115886 @default.
- W3040986370 cites W2810075754 @default.
- W3040986370 cites W2885487916 @default.
- W3040986370 cites W2916118939 @default.
- W3040986370 cites W2922390595 @default.
- W3040986370 cites W2949117887 @default.
- W3040986370 cites W2955425717 @default.
- W3040986370 cites W2962767366 @default.
- W3040986370 cites W2962835968 @default.
- W3040986370 cites W2963125010 @default.
- W3040986370 cites W2963446712 @default.
- W3040986370 cites W2963477586 @default.
- W3040986370 cites W2963521729 @default.
- W3040986370 cites W2963858333 @default.
- W3040986370 cites W2964015378 @default.
- W3040986370 cites W2964081807 @default.
- W3040986370 cites W2965658867 @default.
- W3040986370 cites W2970273366 @default.
- W3040986370 cites W2970482593 @default.
- W3040986370 cites W2970967090 @default.
- W3040986370 cites W2981985696 @default.
- W3040986370 cites W3034733718 @default.
- W3040986370 cites W3035421247 @default.
- W3040986370 cites W3098817529 @default.
- W3040986370 cites W3118608800 @default.
- W3040986370 cites W3143219376 @default.
- W3040986370 doi "https://doi.org/10.48550/arxiv.2007.06559" @default.
- W3040986370 hasPublicationYear "2020" @default.
- W3040986370 type Work @default.
- W3040986370 sameAs 3040986370 @default.
- W3040986370 citedByCount "12" @default.
- W3040986370 countsByYear W30409863702020 @default.
- W3040986370 countsByYear W30409863702021 @default.
- W3040986370 crossrefType "posted-content" @default.
- W3040986370 hasAuthorship W3040986370A5003491365 @default.
- W3040986370 hasAuthorship W3040986370A5024785349 @default.
- W3040986370 hasAuthorship W3040986370A5049098969 @default.
- W3040986370 hasAuthorship W3040986370A5091272738 @default.
- W3040986370 hasBestOaLocation W30409863701 @default.
- W3040986370 hasConcept C11413529 @default.
- W3040986370 hasConcept C132525143 @default.
- W3040986370 hasConcept C154945302 @default.
- W3040986370 hasConcept C175202392 @default.
- W3040986370 hasConcept C22047676 @default.
- W3040986370 hasConcept C41008148 @default.
- W3040986370 hasConcept C45374587 @default.
- W3040986370 hasConcept C50644808 @default.
- W3040986370 hasConcept C73555534 @default.
- W3040986370 hasConcept C80444323 @default.
- W3040986370 hasConcept C86582703 @default.
- W3040986370 hasConceptScore W3040986370C11413529 @default.
- W3040986370 hasConceptScore W3040986370C132525143 @default.
- W3040986370 hasConceptScore W3040986370C154945302 @default.
- W3040986370 hasConceptScore W3040986370C175202392 @default.
- W3040986370 hasConceptScore W3040986370C22047676 @default.
- W3040986370 hasConceptScore W3040986370C41008148 @default.
- W3040986370 hasConceptScore W3040986370C45374587 @default.
- W3040986370 hasConceptScore W3040986370C50644808 @default.
- W3040986370 hasConceptScore W3040986370C73555534 @default.
- W3040986370 hasConceptScore W3040986370C80444323 @default.
- W3040986370 hasConceptScore W3040986370C86582703 @default.
- W3040986370 hasLocation W30409863701 @default.
- W3040986370 hasOpenAccess W3040986370 @default.
- W3040986370 hasPrimaryLocation W30409863701 @default.
- W3040986370 hasRelatedWork W1584270863 @default.
- W3040986370 hasRelatedWork W1595652908 @default.
- W3040986370 hasRelatedWork W2085961337 @default.
- W3040986370 hasRelatedWork W2089093251 @default.
- W3040986370 hasRelatedWork W2106547445 @default.
- W3040986370 hasRelatedWork W2367207292 @default.
- W3040986370 hasRelatedWork W2378845890 @default.
- W3040986370 hasRelatedWork W2386387936 @default.
- W3040986370 hasRelatedWork W2533573066 @default.
- W3040986370 hasRelatedWork W651238688 @default.