Name
Abstract | ||
|---|---|---|
In this paper, we propose a novel edge-labeling graph neural network (EGNN), which adapts a deep neural network on the edge-labeling graph, for few-shot learning. The previous graph neural network (GNN) approaches in few-shot learning have been based on the node-labeling framework, which implicitly models the intra-cluster similarity and the inter-cluster dissimilarity. In contrast, the proposed EGNN learns to predict the edge-labels rather than the node-labels on the graph that enables the evolution of an explicit clustering by iteratively updating the edge-labels with direct exploitation of both intra-cluster similarity and the inter-cluster dissimilarity. It is also well suited for performing on various numbers of classes without retraining, and can be easily extended to perform a transductive inference. The parameters of the EGNN are learned by episodic training with an edge-labeling loss to obtain a well-generalizable model for unseen low-data problem. On both of the supervised and semi-supervised few-shot image classification tasks with two benchmark datasets, the proposed EGNN significantly improves the performances over the existing GNNs. |
Year | DOI | Venue |
|---|---|---|
2019 | 10.1109/CVPR.2019.00010 | 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) |
Keywords | DocType | Volume |
Deep Learning,Recognition: Detection,Categorization,Retrieval,Representation Learning,Segmentation,Grouping a | Conference | abs/1905.01436 |
ISSN | ISBN | Citations |
1063-6919 | 978-1-7281-3294-5 | 17 |
PageRank | References | Authors |
0.69 | 5 | 4 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Kim, Jongmin | 1 | 27 | 2.00 |
| Taesup Kim | 2 | 69 | 5.39 |
| Sungwoong Kim | 3 | 76 | 5.10 |
| Chang D. Yoo | 4 | 375 | 45.88 |