SEGA: Semantic Guided Attention on Visual Prototype for Few-Shot Learning

@article{Yang2021SEGASG,
  title={SEGA: Semantic Guided Attention on Visual Prototype for Few-Shot Learning},
  author={Fengyuan Yang and Ruiping Wang and Xilin Chen},
  journal={2022 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)},
  year={2021},
  pages={1586-1596}
}
Teaching machines to recognize a new category based on few training samples especially only one remains challenging owing to the incomprehensive understanding of the novel category caused by the lack of data. However, human can learn new classes quickly even given few samples since human can tell what discriminative features should be focused on about each category based on both the visual and semantic prior knowledge. To better utilize those prior knowledge, we propose the SEmantic Guided… 
[PDF] Semantic Reader
7 Citations
Background Citations
2
Methods Citations
2

Figures and Tables from this paper

7 Citations

Function-words Adaptively Enhanced Attention Networks for Few-Shot Inverse Relation Classification

A function words adaptively enhanced attention framework (FAEA) for few-shot inverse relation classification, in which a hybrid attention model is designed to attend class-related function words based on meta-learning.

Function-words Enhanced Attention Networks for Few-Shot Inverse Relation Classification

A function words adaptively enhanced attention framework (FAEA) for few-shot inverse relation classification, in which a hybrid attention model is designed to attend class-related function words based on meta-learning.

DiffAlign : Few-shot learning using diffusion based synthesis and alignment

This work proposes DiffAlign which focuses on generating images from class labels that can generate realistic images from texts and employs a maximum mean discrepancy (MMD) loss to align the synthetic images to the real images minimizing the domain gap.

MFHI: Taking Modality-Free Human Identification as Zero-Shot Learning

An initial attempt is taken to formulate a novel Modality-Free Human Identification task as a generic zero-shot learning model in a scalable way, capable of bridging the visual and semantic modalities by learning a discriminative prototype of each identity.

Disentangled Generation with Information Bottleneck for Few-Shot Learning

A novel Information Bottleneck (IB) based Disentangled Generation Framework for FSL, termed as DisGenIB, that can simultaneously guarantee the discrimination and diversity of generated samples and can effectively utilize priors to further facilitate disentanglement.

Few-Shot Meta Learning for Recognizing Facial Phenotypes of Genetic Disorders

A facial recognition model trained on a large corpus of healthy individuals as a pre-task and transferred it to facial phenotype recognition is used and the CNN baseline surpasses previous works, and few-shot meta-learning strate-gies improve retrieval performance in frequent and rare classes.

Knowledge transfer based hierarchical few-shot learning via tree-structured knowledge graph

A knowledge transfer based hierarchical few-shot learning model, which takes advantage of a tree-structured knowledge graph to facilitate the classification results and outperforms several state-of-the-art models.

References

SHOWING 1-10 OF 64 REFERENCES

Few-Shot Image Recognition With Knowledge Transfer

A novel Knowledge Transfer Network architecture (KTN) for few-shot image recognition that jointly incorporates visual feature learning, knowledge inferring and classifier learning into one unified framework for their optimal compatibility.

Knowledge Graph Transfer Network for Few-Shot Recognition

This work represents the semantic correlations in the form of structured knowledge graph and integrates this graph into deep neural networks to promote few-shot learning by a novel Knowledge Graph Transfer Network (KGTN).

Dynamic Few-Shot Visual Learning Without Forgetting

This work proposes to extend an object recognition system with an attention based few-shot classification weight generator, and to redesign the classifier of a ConvNet model as the cosine similarity function between feature representations and classification weight vectors.

Multi-Level Semantic Feature Augmentation for One-Shot Learning

This paper proposes a novel approach to one-shot learning that learns to map a novel sample instance to a concept, relates that concept to the existing ones in the concept space and, using these relationships, generates new instances by interpolating among the concepts to help learning.

Baby steps towards few-shot learning with multiple semantics

Prototype Completion with Primitive Knowledge for Few-Shot Learning

A novel prototype completion based meta-learning framework that first introduces primitive knowledge and extracts representative attribute features as priors, and designs a prototype completion network to learn to complete prototypes with these priors.

Adaptive Cross-Modal Few-Shot Learning

This paper proposes a mechanism that can adaptively combine information from both modalities according to new image categories to be learned and shows that by this adaptive combination of the two modalities, this model outperforms current uni-modality few-shot learning methods and modality-alignment methods by a large margin on all benchmarks and few- shot scenarios tested.

Cross Attention Network for Few-shot Classification

A novel Cross Attention Network is introduced to deal with the problem of unseen classes and a transductive inference algorithm is proposed to alleviate the low-data problem, which iteratively utilizes the unlabeled query set to augment the support set, thereby making the class features more representative.

Learning to Compare: Relation Network for Few-Shot Learning

A conceptually simple, flexible, and general framework for few-shot learning, where a classifier must learn to recognise new classes given only few examples from each, which is easily extended to zero- shot learning.

Rethinking Few-Shot Image Classification: a Good Embedding Is All You Need?

It is shown that a simple baseline: learning a supervised or self-supervised representation on the meta-training set, followed by training a linear classifier on top of this representation, outperforms state-of-the-art few-shot learning methods.
...