Matches in SemOpenAlex for { <https://semopenalex.org/work/W4383469084> ?p ?o ?g. }
Showing items 1 to 73 of
73
with 100 items per page.
- W4383469084 abstract "Diffusion models based on permutation-equivariant networks can learn permutation-invariant distributions for graph data. However, in comparison to their non-invariant counterparts, we have found that these invariant models encounter greater learning challenges since 1) their effective target distributions exhibit more modes; 2) their optimal one-step denoising scores are the score functions of Gaussian mixtures with more components. Motivated by this analysis, we propose a non-invariant diffusion model, called $textit{SwinGNN}$, which employs an efficient edge-to-edge 2-WL message passing network and utilizes shifted window based self-attention inspired by SwinTransformers. Further, through systematic ablations, we identify several critical training and sampling techniques that significantly improve the sample quality of graph generation. At last, we introduce a simple post-processing trick, $textit{i.e.}$, randomly permuting the generated graphs, which provably converts any graph generative model to a permutation-invariant one. Extensive experiments on synthetic and real-world protein and molecule datasets show that our SwinGNN achieves state-of-the-art performances. Our code is released at https://github.com/qiyan98/SwinGNN." @default.
- W4383469084 created "2023-07-07" @default.
- W4383469084 creator A5040875615 @default.
- W4383469084 creator A5041125507 @default.
- W4383469084 creator A5048686150 @default.
- W4383469084 creator A5053138553 @default.
- W4383469084 creator A5066401817 @default.
- W4383469084 date "2023-07-04" @default.
- W4383469084 modified "2023-09-23" @default.
- W4383469084 title "SwinGNN: Rethinking Permutation Invariance in Diffusion Models for Graph Generation" @default.
- W4383469084 doi "https://doi.org/10.48550/arxiv.2307.01646" @default.
- W4383469084 hasPublicationYear "2023" @default.
- W4383469084 type Work @default.
- W4383469084 citedByCount "0" @default.
- W4383469084 crossrefType "posted-content" @default.
- W4383469084 hasAuthorship W4383469084A5040875615 @default.
- W4383469084 hasAuthorship W4383469084A5041125507 @default.
- W4383469084 hasAuthorship W4383469084A5048686150 @default.
- W4383469084 hasAuthorship W4383469084A5053138553 @default.
- W4383469084 hasAuthorship W4383469084A5066401817 @default.
- W4383469084 hasBestOaLocation W43834690841 @default.
- W4383469084 hasConcept C11413529 @default.
- W4383469084 hasConcept C118615104 @default.
- W4383469084 hasConcept C121332964 @default.
- W4383469084 hasConcept C132525143 @default.
- W4383469084 hasConcept C154945302 @default.
- W4383469084 hasConcept C163716315 @default.
- W4383469084 hasConcept C167966045 @default.
- W4383469084 hasConcept C171036898 @default.
- W4383469084 hasConcept C190470478 @default.
- W4383469084 hasConcept C202444582 @default.
- W4383469084 hasConcept C21308566 @default.
- W4383469084 hasConcept C24890656 @default.
- W4383469084 hasConcept C33923547 @default.
- W4383469084 hasConcept C37914503 @default.
- W4383469084 hasConcept C39890363 @default.
- W4383469084 hasConcept C41008148 @default.
- W4383469084 hasConcept C62520636 @default.
- W4383469084 hasConcept C80444323 @default.
- W4383469084 hasConceptScore W4383469084C11413529 @default.
- W4383469084 hasConceptScore W4383469084C118615104 @default.
- W4383469084 hasConceptScore W4383469084C121332964 @default.
- W4383469084 hasConceptScore W4383469084C132525143 @default.
- W4383469084 hasConceptScore W4383469084C154945302 @default.
- W4383469084 hasConceptScore W4383469084C163716315 @default.
- W4383469084 hasConceptScore W4383469084C167966045 @default.
- W4383469084 hasConceptScore W4383469084C171036898 @default.
- W4383469084 hasConceptScore W4383469084C190470478 @default.
- W4383469084 hasConceptScore W4383469084C202444582 @default.
- W4383469084 hasConceptScore W4383469084C21308566 @default.
- W4383469084 hasConceptScore W4383469084C24890656 @default.
- W4383469084 hasConceptScore W4383469084C33923547 @default.
- W4383469084 hasConceptScore W4383469084C37914503 @default.
- W4383469084 hasConceptScore W4383469084C39890363 @default.
- W4383469084 hasConceptScore W4383469084C41008148 @default.
- W4383469084 hasConceptScore W4383469084C62520636 @default.
- W4383469084 hasConceptScore W4383469084C80444323 @default.
- W4383469084 hasLocation W43834690841 @default.
- W4383469084 hasOpenAccess W4383469084 @default.
- W4383469084 hasPrimaryLocation W43834690841 @default.
- W4383469084 hasRelatedWork W1997059631 @default.
- W4383469084 hasRelatedWork W2027233917 @default.
- W4383469084 hasRelatedWork W2093536585 @default.
- W4383469084 hasRelatedWork W3154973012 @default.
- W4383469084 hasRelatedWork W3210824699 @default.
- W4383469084 hasRelatedWork W4200633419 @default.
- W4383469084 hasRelatedWork W4226058601 @default.
- W4383469084 hasRelatedWork W4287126906 @default.
- W4383469084 hasRelatedWork W4383469084 @default.
- W4383469084 hasRelatedWork W3008759432 @default.
- W4383469084 isParatext "false" @default.
- W4383469084 isRetracted "false" @default.
- W4383469084 workType "article" @default.