Knowledge Graph Transfer Network for Few-Shot Recognition

Chen, Riquan; Chen, Tianshui; Hui, Xiaolu; Wu, Hefeng; Li, Guanbin; Li, Lin

doi:10.1609/aaai.v34i07.6630

Cited by 48 publications

(47 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Variational scaling [62,2020] CNAPS [63,2020] TEAM [65,2020] SEN [66,2020] DeepEMD [68,2020] FSL with embedded class models [54,2019] Two-stage FSL [55,2020] TapNet [56,2019] FSL with global class representations [57,2019] KGTN [58,2020] Learning task-agnostic features Siamese Network [13,2015] Matching Network [9,2016] TPN [29,2019] PTN [30,2021] DN4 [32,2019] ATL-Net [35,2020] COMET [36,2021] PARN [33,2019] Attentive Prototype [37,2020] Cross-domain FSL [38,2020]…”

Section: Few-shot Deep Metric Learning Methodsmentioning

confidence: 99%

Revisiting metric learning for few-shot image classification

et al. 2020

Neurocomputing

View full text Add to dashboard Cite

The goal of few-shot learning is to recognize new visual concepts with just a few amount of labeled samples in each class. Recent effective metric-based few-shot approaches employ neural networks to learn a feature similarity comparison between query and support examples. However, the importance of feature embedding, i.e., exploring the relationship among training samples, is neglected. In this work, we present a simple yet powerful baseline for few-shot classification by emphasizing the importance of feature embedding. Specifically, we revisit the classical triplet network from deep metric learning, and extend it into a deep K-tuplet network for few-shot learning, utilizing the relationship among the input samples to learn a general representation learning via episode-training. Once trained, our network is able to extract discriminative features for unseen novel categories and can be seamlessly incorporated with a non-linear distance metric function to facilitate the few-shot classification. Our result on the miniImageNet benchmark outperforms other metric-based few-shot classification methods. More importantly, when evaluated on completely different datasets (Caltech-101, CUB-200, Stanford Dogs and Cars) using the model trained with miniImageNet, our method significantly outperforms prior methods, demonstrating its superior capability to generalize to unseen classes.

show abstract

Section: Few-shot Deep Metric Learning Methodsmentioning

confidence: 99%

Revisiting metric learning for few-shot image classification

et al. 2020

Neurocomputing

View full text Add to dashboard Cite

show abstract

“…Since we have extracted CNN representations for each pedestrian image, they remain characteristics of their respective view, where causes huge distribution-gap and view-interference for unsupervised identity matching. Hence, this paper introduces graph convolutional layers to alleviate the distribution-gap by the adjacent path of intra-view and interview graph A, inspired by [1], [4], [39].…”

Section: B Unsupervised Person Re-idmentioning

confidence: 99%

“…The main contributions of our work can be summarized as follows: (1) this paper proposes a cross-view graph adversarial network for solving the view-inference in unsupervised manner, and losses of reconstruction, adversarial learning and identity-preserving provide training directions for unsupervised pedestrian feature learning; (2) We design a multi-order discriminative feature learning module to estimate triplet samples from nearest neighbors following multiple cross-view orders, that can exploit more meaningful results reveal that cross-view graph adversarial network and exploiting multi-order relationships significantly improve the performance on unsupervised person re-id, and our method outperforms existing unsupervised re-id methods.…”

mentioning

confidence: 99%

“…Though CNN network represents pedestrian images by M p and M g , feature maps follow each view-biased distribution and matching identities in this feature space is severely interfered by the large cross-view distribution-gap. Recently, Graph Convolutional Network (GCN)[12] has widely adopted to mine mutual correlations among feature vectors, and it is an effective mean for addressing various featurelevel distribution gaps[1],[4],[39]. Inspired by these works, we propose a cross-view graph convolutional network to bridging distribution-gap between M p and M g , without any help of identity annotations.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Unsupervised Person Re-Identification via Multi-Order Cross-View Graph Adversarial Network

Lai

2021

IEEE Access

View full text Add to dashboard Cite

“…Microsoft-COCO contains only 80 categories, but current works [73], [74] for few-shot learning may contain thou- Comparison of the mAP (in %) on 1-shot and 5-shot settings on the VG-500 dataset. We present the results of 1-shot and 5-shot settings.…”

Section: Comparison On Visual Genome 500mentioning

confidence: 99%

Multi-label Few-shot Learning for Sound Event Recognition

Cheng

Chou

Yang

2019

2019 IEEE 21st International Workshop on Multimedia Signal Processing (MMSP)

View full text Add to dashboard Cite

Recognizing multiple labels of an image is a practical yet challenging task, and remarkable progress has been achieved by searching for semantic regions and exploiting label dependencies. However, current works utilize RNN/LSTM to implicitly capture sequential region/label dependencies, which cannot fully explore mutual interactions among the semantic regions/labels and do not explicitly integrate label co-occurrences. In addition, these works require large amounts of training samples for each category, and they are unable to generalize to novel categories with limited samples. To address these issues, we propose a knowledge-guided graph routing (KGGR) framework, which unifies prior knowledge of statistical label correlations with deep neural networks. The framework exploits prior knowledge to guide adaptive information propagation among different categories to facilitate multi-label analysis and reduce the dependency of training samples. Specifically, it first builds a structured knowledge graph to correlate different labels based on statistical label co-occurrence. Then, it introduces the label semantics to guide learning semantic-specific features to initialize the graph, and it exploits a graph propagation network to explore graph node interactions, enabling learning contextualized image feature representations. Moreover, we initialize each graph node with the classifier weights for the corresponding label and apply another propagation network to transfer node messages through the graph. In this way, it can facilitate exploiting the information of correlated labels to help train better classifiers, especially for labels with limited training samples. We conduct extensive experiments on the traditional multi-label image recognition (MLR) and multi-label few-shot learning (ML-FSL) tasks and show that our KGGR framework outperforms the current state-of-the-art methods by sizable margins on the public benchmarks.

show abstract

Knowledge Graph Transfer Network for Few-Shot Recognition

Cited by 48 publications

References 29 publications

Revisiting metric learning for few-shot image classification

Revisiting metric learning for few-shot image classification

Unsupervised Person Re-Identification via Multi-Order Cross-View Graph Adversarial Network

Multi-label Few-shot Learning for Sound Event Recognition

Contact Info

Product

Resources

About