Behavior regularized prototypical networks for semi-supervised few-shot image classification

Huang, Shixin; Zeng, Xianlin; Wu, Si; Yu, Zhiwen; Azzam, Mohamed; Wong, Hau-San

doi:10.1016/j.patcog.2020.107765

Cited by 20 publications

(13 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition, we set a supervised experiment including the ensemble model Ho-ETPN and He-ETPN, and a semi-supervised experiment including the setting of distractor classes. These approaches are particularly divided into optimization-based (MAML [22]), ensemble-based (EBDM-Euc [38], HGNN [39], E 3 BM+MAML [40]), graph-based (TPN [25], EPNet [27], TPRN [31], DSN [32], EGNN [33], PRWN [35], GNN [52], BGNN * [53], DPGN * [54]), and metricbased (MatchingNet [8], Proto Net [9], TADAM [13], BR-ProtoNet [36], SSFormers [55], CGRN [56], HMRN [57]) approaches. Moreover, we conduct 5-way 1-shot and 5-shot experiments, which are standard few-shot learning settings.…”

Section: Methodsmentioning

confidence: 99%

“…PRWN proposed prototypical random walk networks to promote prototypical magnetization of the learning representation [35]. BR-ProtoNet exploited unlabeled data and constructed complementary constraints to learn a generalizable metric [36]. In this paper, we adopt transductive inference to utilize unlabeled data and distractor classes irrelevant to the classification task to boost robustness against perturbations.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Ensemble Transductive Propagation Network for Semi-Supervised Few-Shot Learning

Pan,

Li,

Zheng

2024

Entropy

View full text Add to dashboard Cite

Few-shot learning aims to solve the difficulty in obtaining training samples, leading to high variance, high bias, and over-fitting. Recently, graph-based transductive few-shot learning approaches supplement the deficiency of label information via unlabeled data to make a joint prediction, which has become a new research hotspot. Therefore, in this paper, we propose a novel ensemble semi-supervised few-shot learning strategy via transductive network and Dempster–Shafer (D-S) evidence fusion, named ensemble transductive propagation networks (ETPN). First, we present homogeneity and heterogeneity ensemble transductive propagation networks to better use the unlabeled data, which introduce a preset weight coefficient and provide the process of iterative inferences during transductive propagation learning. Then, we combine the information entropy to improve the D-S evidence fusion method, which improves the stability of multi-model results fusion from the pre-processing of the evidence source. Third, we combine the L2 norm to improve an ensemble pruning approach to select individual learners with higher accuracy to participate in the integration of the few-shot model results. Moreover, interference sets are introduced to semi-supervised training to improve the anti-disturbance ability of the mode. Eventually, experiments indicate that the proposed approaches outperform the state-of-the-art few-shot model. The best accuracy of ETPN increases by 0.3% and 0.28% in the 5-way 5-shot, and by 3.43% and 7.6% in the 5-way 1-shot on miniImagNet and tieredImageNet, respectively.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Ensemble Transductive Propagation Network for Semi-Supervised Few-Shot Learning

Pan,

Li,

Zheng

2024

Entropy

View full text Add to dashboard Cite

show abstract

“…Few-shot learning aims to recognize novel categories utilizing a few training examples. To solve this problem, most researchers have focused on few-shot classification where an object is clearly placed in an image [24,25,26]. In other words, the approaches do not consider the concept of backgrounds and bounding box regression.…”

Section: Few-shot Object Detectionmentioning

confidence: 99%

Spatial reasoning for few-shot object detection

Kim

Jung

Lee

2021

Pattern Recognition

View full text Add to dashboard Cite

“…In both works, a class prediction is made based on the similarity of an encoded input to the model's prototypes via a simple distance metric. Lastly, especially in the context of few-shot learning, prototypes show advantageous properties [74,62,60,35]. Disentanglement.…”

Section: Related Workmentioning

confidence: 99%

Interactive Disentanglement: Learning Concepts by Interacting with their Prototype Representations

Stammer¹,

Memmel²,

Schramowski³

et al. 2021

Preprint

View full text Add to dashboard Cite

Learning visual concepts from raw images without strong supervision is a challenging task. In this work, we show the advantages of prototype representations for understanding and revising the latent space of neural concept learners. For this purpose, we introduce interactive Concept Swapping Networks (iCSNs), a novel framework for learning conceptgrounded representations via weak supervision and implicit prototype representations. iCSNs learn to bind conceptual information to specific prototype slots by swapping the latent representations of paired images. This semantically grounded and discrete latent space facilitates human understanding and human-machine interaction. We support this claim by conducting experiments on our novel data set "Elementary Concept Reasoning" (ECR), focusing on visual concepts shared by geometric objects.

show abstract

Behavior regularized prototypical networks for semi-supervised few-shot image classification

Cited by 20 publications

References 8 publications

Ensemble Transductive Propagation Network for Semi-Supervised Few-Shot Learning

Ensemble Transductive Propagation Network for Semi-Supervised Few-Shot Learning

Spatial reasoning for few-shot object detection

Interactive Disentanglement: Learning Concepts by Interacting with their Prototype Representations

Contact Info

Product

Resources

About