• Corpus ID: 236493217

FREE: Feature Refinement for Generalized Zero-Shot Learning

@article{Chen2021FREEFR,
  title={FREE: Feature Refinement for Generalized Zero-Shot Learning},
  author={Shiming Chen and Wenjie Wang and Beihao Xia and Qinmu Peng and Xinge You and Feng Zheng and Ling Shao},
  journal={ArXiv},
  year={2021},
  volume={abs/2107.13807}
}
Generalized zero-shot learning (GZSL) has achieved significant progress, with many efforts dedicated to overcoming the problems of visual-semantic domain gap and seenunseen bias. However, most existing methods directly use feature extraction models trained on ImageNet alone, ignoring the cross-dataset bias between ImageNet and GZSL benchmarks. Such a bias inevitably results in poor-quality visual features for GZSL tasks, which potentially limits the recognition performance on both seen and… 

Figures and Tables from this paper

TransZero++: Cross Attribute-Guided Transformer for Zero-Shot Learning
TLDR
A cross attribute-guided Transformer network to refine visual features and learn accurate attribute localization for semantic-augmented visual embedding representations in ZSL, which achieves the new state-of-the-art results on three golden and challenging ZSL benchmarks.
TransZero: Attribute-guided Transformer for Zero-Shot Learning
TLDR
This paper proposes an attribute-guided Transformer network, termed TransZero, to refine visual features and learn attribute localization for discriminative visual embedding representations in ZSL, and achieves the new state of the art on three ZSL benchmarks.
HSVA: Hierarchical Semantic-Visual Adaptation for Zero-Shot Learning
TLDR
A supervised adversarial discrepancy (SAD) module is proposed to adversarially minimize the discrepancy between the predictions of two task-specific classifiers, thus making the visual and semantic feature manifolds more closely aligned.
Robust Region Feature Synthesizer for Zero-Shot Object Detection
TLDR
The core challenges in this research area are revealed: how to synthesize robust region features that are as intra-class diverse and inter-class separable as the real samples, so that strong unseen object detectors can be trained upon them and how to avoid miss-classifying the real unseen objects as image backgrounds.
Imaginative Walks: Generative Random Walk Deviation Loss for Improved Unseen Learning Representation
TLDR
A novel loss for generative models, dubbed as GRaWD (Generative Random Walk Deviation), to improve learning representations of unexplored visual spaces, which can improve StyleGAN1 and StyleGAN2 generation quality and create novel art that is significantly more preferred.
MFHI: Taking Modality-free Human Identification as Zero-shot Learning
  • Zhizhe Liu, Xingxing Zhang, Zhenfeng Zhu, Shuai Zheng, Yao Zhao, Jian Cheng
  • Computer Science
    IEEE Transactions on Circuits and Systems for Video Technology
  • 2021
TLDR
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.
Model Adaptation: Historical Contrastive Learning for Unsupervised Domain Adaptation without Source Data
TLDR
An innovative historical contrastive learning (HCL) technique that exploits historical source hypothesis to make up for the absence of source data in unsupervised model adaptation (UMA).

References

SHOWING 1-10 OF 73 REFERENCES
Domain-Aware Visual Bias Eliminating for Generalized Zero-Shot Learning
TLDR
A novel Domain-aware Visual Bias Eliminating (DVBE) network is proposed that constructs two complementary visual representations, i.e., semantic-free and semantic-aligned, to treat seen and unseen domains separately and outperforms existing methods by averaged 5.7% improvement.
Rethinking Zero-Shot Learning: A Conditional Visual Classification Perspective
TLDR
This work reformulates ZSL as a conditioned visual classification problem, i.e., classifying visual features based on the classifiers learned from the semantic descriptions, and develops algorithms targeting various ZSL settings.
Self-Supervised Domain-Aware Generative Network for Generalized Zero-Shot Learning
TLDR
This is the first work to introduce self-supervised learning into GZSL as a learning guidance and designs a cross-domain feature generating module to synthesize samples with high fidelity based on class embeddings, which involves a novel target domain discriminator to preserve the domain consistency.
Transferable Contrastive Network for Generalized Zero-Shot Learning
TLDR
A novel Transferable Contrastive Network (TCN) is proposed that explicitly transfers knowledge from the source classes to the target classes, and is more robust to recognize the target images.
Graph and Autoencoder Based Feature Extraction for Zero-shot Learning
TLDR
A novel framework named Graph and Autoencoder Based Feature Extraction (GAFE) is formulated to seek a low-rank mapping to preserve the sub-manifold of samples in the embedding space and a graph is constructed to guarantee the learned mapping can preserve the local intrinsic structure of the data.
Marginalized Latent Semantic Encoder for Zero-Shot Learning
  • Zhengming Ding, Hongfu Liu
  • Computer Science
    2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
  • 2019
TLDR
This paper designs a Marginalized Latent Semantic Encoder (MLSE), which is learned on the augmented seen visual features and the latent semantic representation, and whose latent semantics are discovered under an adaptive graph reconstruction scheme based on the provided semantics.
Generalized Zero- and Few-Shot Learning via Aligned Variational Autoencoders
TLDR
This work proposes a model where a shared latent space of image features and class embeddings is learned by modality-specific aligned variational autoencoders, and align the distributions learned from images and from side-information to construct latent features that contain the essential multi-modal information associated with unseen classes.
Generalized Zero-Shot Learning with Deep Calibration Network
TLDR
This paper proposes a novel Deep Calibration Network (DCN) approach towards this generalized zero-shot learning paradigm, which enables simultaneous calibration of deep networks on the confidence of source classes and uncertainty of target classes.
A Boundary Based Out-of-Distribution Classifier for Generalized Zero-Shot Learning
TLDR
This paper proposes a boundary based Out-of-Distribution (OOD) classifier which classifies the unseen and seen domains by only using seen samples for training and extensively validate the approach on five popular benchmark datasets including AWA1, AWA2, CUB, FLO and SUN.
Attentive Region Embedding Network for Zero-Shot Learning
  • Guosen Xie, L. Liu, +5 authors L. Shao
  • Computer Science
    2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
  • 2019
TLDR
To discover (semantic) regions, the attentive region embedding network (AREN) is proposed, which is tailored to advance the ZSL task and achieves state-of-the-art performances under ZSLSetting, and compelling results under generalized ZSL setting.
...
1
2
3
4
5
...