Few-Shot Learning with Graph Neural Networks

Moreno, Víctor; Bruna, Joan

doi:10.48550/arxiv.1711.04043

Cited by 148 publications

(200 citation statements)

References 15 publications

Supporting

Mentioning

199

Contrasting

Order By: Relevance

“…For comparisons, different approaches were used for fault conditions with the same dataset, which are Siamese Net, Matching Net, SAE+RF [34,36]. Among them, Siamese Net is recognized as a simple and effective method, which is widely used in computer science, finance, medicine and other fields.…”

Section: Resultsmentioning

confidence: 99%

Exploiting graph neural network with one-shot learning for fault diagnosis of rotating machinery (PHM2021)

Yang

Bai

et al. 2022

Preprint

View full text Add to dashboard Cite

Due to lack of training samples, overfitting is a severe problem in fault diagnosis for mechanical devices, especially for rotating machinery. In this paper, a graph neural network (GNN) method with one-shot learning is proposed for fault diagnosis of rotating machinery. Convolutional Neural Network (CNN) is applied to extract the feature vectors and generate codes for one-shot learning. With adjacency matrix in GNN, the proposed method can achieve fault classification for rotating machinery with small dataset. Rotate vector (RV) reducer of the industrial robot and bearing of the rotating shaft were chosen as experimental subjects. Experimental results show the high accuracy of classification in both experiments with the proposed method. To further verify the efficiency of this method, Siamese Net, Matching Net and SAE+RF were chosen as the comparisons. The results indicate the proposed method outperforms all the selected methods for fault classification in both rotating machineries.

show abstract

Section: Resultsmentioning

confidence: 99%

Exploiting graph neural network with one-shot learning for fault diagnosis of rotating machinery (PHM2021)

Yang

Bai

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Initialization based methods aim to learn good model initialization (i.e., the parameters of a network) so that the classifier for novel classes can be learned with a few labeled samples and a few gradient updated steps [38,8,9,42]. Metric learning based methods aim to learn a sophisticated comparison model to determine the similarity of two images [32,48,46,47,35,12]. Hallucination based methods learn a generator from samples in the base classes and use the learned generator to hallucinate new novel class samples for data augmentation [14,43,55,5,4].…”

Section: Few Shot Learningmentioning

confidence: 99%

Fine-Grained Few Shot Learning with Foreground Object Transformation

Wang¹,

Song²,

Yang³

et al. 2021

Preprint

View full text Add to dashboard Cite

Traditional fine-grained image classification generally requires abundant labeled samples to deal with the low inter-class variance but high intra-class variance problem. However, in many scenarios we may have limited samples for some novel sub-categories, leading to the fine-grained few shot learning (FG-FSL) setting. To address this challenging task, we propose a novel method named foreground object transformation (FOT), which is composed of a foreground object extractor and a posture transformation generator. The former aims to remove image background, which tends to increase the difficulty of fine-grained image classification as it amplifies the intra-class variance while reduces inter-class variance. The latter transforms the posture of the foreground object to generate additional samples for the novel sub-category. As a data augmentation method, FOT can be conveniently applied to any existing few shot learning algorithm and greatly improve its performance on FG-FSL tasks. In particular, in combination with FOT, simple fine-tuning baseline methods can be competitive with the state-of-the-art methods both in inductive setting and transductive setting. Moreover, FOT can further boost the performances of latest excellent methods and bring them up to the new state-of-the-art. In addition, we also show the effectiveness of FOT on general FSL tasks.

show abstract

“…As shown in Figure 2, our framework integrates an object-level GCN and an image-level GCN for visual feature encoding. The first GCN focuses on objects in local regions to explore spatial objectlevel relation [6,39,53,61], while the second one targets on the image-level similarity relation [12,47] amongst multiple similar images. We calculate the similarity of different images and select the images with high similarity as a meaningful complementary global visual embedding to ensure a more reasonable and accurate text description generation.…”

Section: Introductionmentioning

confidence: 99%

Dual Graph Convolutional Networks with Transformer and Curriculum Learning for Image Captioning

Dong

Long²,

Xu³

et al. 2021

Proceedings of the 29th ACM International Conference on Multimedia

View full text Add to dashboard Cite

Existing image captioning methods just focus on understanding the relationship between objects or instances in a single image, without exploring the contextual correlation existed among contextual image. In this paper, we propose Dual Graph Convolutional Networks (Dual-GCN) with transformer and curriculum learning for image captioning. In particular, we not only use an object-level GCN to capture the object to object spatial relation within a single image, but also adopt an image-level GCN to capture the feature information provided by similar images. With the well-designed Dual-GCN, we can make the linguistic transformer better understand the relationship between different objects in a single image and make full use of similar images as auxiliary information to generate a reasonable caption description for a single image. Meanwhile, with a cross-review strategy introduced to determine difficulty levels, we adopt curriculum learning as the training strategy to increase the robustness and generalization of our proposed model. We conduct extensive experiments on the large-scale MS COCO dataset, and the experimental results powerfully demonstrate that our proposed method outperforms recent state-of-the-art approaches. It achieves a BLEU-1 score of 82.2 and a BLEU-2 score of 67.6. Our source code is available at https:// github.com/ Unbear430/ DGCN-for-image-captioning. CCS CONCEPTS• Computing methodologies → Natural language processing; Scene understanding.

show abstract

Few-Shot Learning with Graph Neural Networks

Cited by 148 publications

References 15 publications

Exploiting graph neural network with one-shot learning for fault diagnosis of rotating machinery (PHM2021)

Exploiting graph neural network with one-shot learning for fault diagnosis of rotating machinery (PHM2021)

Fine-Grained Few Shot Learning with Foreground Object Transformation

Dual Graph Convolutional Networks with Transformer and Curriculum Learning for Image Captioning

Contact Info

Product

Resources

About