Matches in SemOpenAlex for { <https://semopenalex.org/work/W4285856336> ?p ?o ?g. }
Showing items 1 to 69 of
69
with 100 items per page.
- W4285856336 abstract "Graph Convolutional Networks (GCNs) have recently attracted vast interest and achieved state-of-the-art performance on graphs, but its success could typically hinge on careful training with amounts of expensive and time-consuming labeled data. To alleviate labeled data scarcity, self-training methods have been widely adopted on graphs by labeling high-confidence unlabeled nodes and then adding them to the training step. In this line, we empirically make a thorough study for current self-training methods on graphs. Surprisingly, we find that high-confidence unlabeled nodes are not always useful, and even introduce the distribution shift issue between the original labeled dataset and the augmented dataset by self-training, severely hindering the capability of self-training on graphs. To this end, in this paper, we propose a novel Distribution Recovered Graph Self-Training framework (DR-GST), which could recover the distribution of the original labeled dataset. Specifically, we first prove the equality of loss function in self-training framework under the distribution shift case and the population distribution if each pseudo-labeled node is weighted by a proper coefficient. Considering the intractability of the coefficient, we then propose to replace the coefficient with the information gain after observing the same changing trend between them, where information gain is respectively estimated via both dropout variational inference and dropedge variational inference in DR-GST. However, such a weighted loss function will enlarge the impact of incorrect pseudo labels. As a result, we apply the loss correction method to improve the quality of pseudo labels. Both our theoretical analysis and extensive experiments on five benchmark datasets demonstrate the effectiveness of the proposed DR-GST, as well as each well-designed component in DR-GST." @default.
- W4285856336 created "2022-07-20" @default.
- W4285856336 creator A5018805094 @default.
- W4285856336 creator A5028576686 @default.
- W4285856336 creator A5051805730 @default.
- W4285856336 creator A5058010200 @default.
- W4285856336 creator A5062877968 @default.
- W4285856336 creator A5089103033 @default.
- W4285856336 date "2022-01-27" @default.
- W4285856336 modified "2023-09-27" @default.
- W4285856336 title "Confidence May Cheat: Self-Training on Graph Neural Networks under Distribution Shift" @default.
- W4285856336 doi "https://doi.org/10.48550/arxiv.2201.11349" @default.
- W4285856336 hasPublicationYear "2022" @default.
- W4285856336 type Work @default.
- W4285856336 citedByCount "0" @default.
- W4285856336 crossrefType "posted-content" @default.
- W4285856336 hasAuthorship W4285856336A5018805094 @default.
- W4285856336 hasAuthorship W4285856336A5028576686 @default.
- W4285856336 hasAuthorship W4285856336A5051805730 @default.
- W4285856336 hasAuthorship W4285856336A5058010200 @default.
- W4285856336 hasAuthorship W4285856336A5062877968 @default.
- W4285856336 hasAuthorship W4285856336A5089103033 @default.
- W4285856336 hasBestOaLocation W42858563361 @default.
- W4285856336 hasConcept C11413529 @default.
- W4285856336 hasConcept C119857082 @default.
- W4285856336 hasConcept C121332964 @default.
- W4285856336 hasConcept C132525143 @default.
- W4285856336 hasConcept C144024400 @default.
- W4285856336 hasConcept C149923435 @default.
- W4285856336 hasConcept C153294291 @default.
- W4285856336 hasConcept C154945302 @default.
- W4285856336 hasConcept C2776145597 @default.
- W4285856336 hasConcept C2776214188 @default.
- W4285856336 hasConcept C2777211547 @default.
- W4285856336 hasConcept C2908647359 @default.
- W4285856336 hasConcept C41008148 @default.
- W4285856336 hasConcept C51632099 @default.
- W4285856336 hasConcept C80444323 @default.
- W4285856336 hasConceptScore W4285856336C11413529 @default.
- W4285856336 hasConceptScore W4285856336C119857082 @default.
- W4285856336 hasConceptScore W4285856336C121332964 @default.
- W4285856336 hasConceptScore W4285856336C132525143 @default.
- W4285856336 hasConceptScore W4285856336C144024400 @default.
- W4285856336 hasConceptScore W4285856336C149923435 @default.
- W4285856336 hasConceptScore W4285856336C153294291 @default.
- W4285856336 hasConceptScore W4285856336C154945302 @default.
- W4285856336 hasConceptScore W4285856336C2776145597 @default.
- W4285856336 hasConceptScore W4285856336C2776214188 @default.
- W4285856336 hasConceptScore W4285856336C2777211547 @default.
- W4285856336 hasConceptScore W4285856336C2908647359 @default.
- W4285856336 hasConceptScore W4285856336C41008148 @default.
- W4285856336 hasConceptScore W4285856336C51632099 @default.
- W4285856336 hasConceptScore W4285856336C80444323 @default.
- W4285856336 hasLocation W42858563361 @default.
- W4285856336 hasOpenAccess W4285856336 @default.
- W4285856336 hasPrimaryLocation W42858563361 @default.
- W4285856336 hasRelatedWork W167670643 @default.
- W4285856336 hasRelatedWork W2605524926 @default.
- W4285856336 hasRelatedWork W2753840555 @default.
- W4285856336 hasRelatedWork W2799803467 @default.
- W4285856336 hasRelatedWork W2953328427 @default.
- W4285856336 hasRelatedWork W3100814071 @default.
- W4285856336 hasRelatedWork W4210794429 @default.
- W4285856336 hasRelatedWork W4280575929 @default.
- W4285856336 hasRelatedWork W4280592718 @default.
- W4285856336 hasRelatedWork W3112985236 @default.
- W4285856336 isParatext "false" @default.
- W4285856336 isRetracted "false" @default.
- W4285856336 workType "article" @default.