Transductive Multi-View Zero-Shot Learning
Open accessThe DOI matches its Crossref record; 72 of the 74 references it cites were found in a registry or an index.
Abstract
Most existing zero-shot learning approaches exploit transfer learning via an intermediate semantic representation shared between an annotated auxiliary dataset and a target dataset with different classes and no annotation. A projection from a low-level feature space to the semantic representation space is learned from the auxiliary dataset and applied without adaptation to the target dataset. In this paper we identify two inherent limitations with these approaches. First, due to having disjoint and potentially unrelated classes, the projection functions learned from the auxiliary dataset/domain are biased when applied directly to the target dataset/domain. We call this problem the projection domain shift problem and propose a novel framework, transductive multi-view embedding, to solve it. The second limitation is the prototype sparsity problem which refers to the fact that for each target class, only a single prototype is available for zero-shot learning given a semantic representation. To overcome this problem, a novel heterogeneous multi-view hypergraph label propagation method is formulated for zero-shot learning in the transductive embedding space. It effectively exploits the complementary information offered by different semantic representations and takes advantage of the manifold structures of multiple representation spaces in a coherent manner. We demonstrate through extensive experiments that the proposed approach (1) rectifies the projection shift between the auxiliary and target domains, (2) exploits the complementarity of multiple semantic representations, (3) significantly outperforms existing methods for both zero-shot and N-shot recognition on three image and video benchmark datasets, and (4) enables novel cross-view annotation tasks.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
ImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155 · cited 1×
“What is interesting to see here is that Overfeat > Decaf since both are based on the same Convolutional Neural Network (CNN) model of [32].”
BackgroundCollective dynamics of ‘small-world’ networks
Watts, D. J., & Strogatz, S. H. (1998) · Nature · doi:10.1038/30918 · Cited by 43,060
LIBSVM: A library for support vector machines
Chang, C.-C., & Lin, C.-J. (2011) · ACM Transactions on Intelligent Systems and Technology · doi:10.1145/1961189.1961199 · Cited by 40,918 · cited 1×
“However, this does not effectively exploit the multi-view complementarity, and suffers from labelled data (prototype) sparsity.”
Background(2008) · Journal of Machine Learning Research · Cited by 33,051
Pan, S. J., & Yang, Q. (2010) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2009.191 · Cited by 23,387
Learning Deep Architectures for AI
Bengio, Y. (2009) · Foundations and Trends® in Machine Learning · doi:10.1561/2200000006 · Cited by 9,239 · cited 1×
“The advantages of heterogeneous hypergraphs We argue that the pairwise similarity of heterogeneous hypergraph is a distributed representation [3].”
BackgroundRecognition-by-components: A theory of human image understanding.
Biederman, I. (1987) · Psychological Review · doi:10.1037/0033-295x.94.2.115 · Cited by 5,621 · cited 1×
“Humans can distinguish 30,000 basic object classes [3] and many more subordinate ones (e.g. breeds of dogs).”
BackgroundLearning Deep Architectures for AI
Bengio, Y. (2009) · now publishers, Inc. eBooks · doi:10.1561/9781601982957 · Cited by 4,941
Semi-Supervised Learning Literature Survey
Zhu, X. (2005) · Minds at UW (University of Wisconsin) · Cited by 4,505
DeCAF: A Deep Convolutional Activation Feature for Generic Visual Recognition
Donahue, J., Jia, Y., Vinyals, O., Hoffman, J., Zhang, N., Tzeng, E., & Darrell, T. (2013) · arXiv · doi:10.48550/arxiv.1310.1531 · Cited by 4,019 · cited 1×
“In addition, the Decaf (D) [10] feature is also considered for AwA.”
Background
Research works printed but not linked: 5
- [39] E. Rosch. Classification of real-world objects: Origins and representations in cognition. Thinking: Readings in Cognitive Science , 1977. -
- [32] T. Mikolov, K. Chen, G. Corrado, and J. Dean. Efficient estimation of word representation in vector space. In Proceedings of Workshop at ICLR , 2013. -
- [33] M. Palatucci, G. Hinton, D. Pomerleau, and T. M. Mitchell. Zeroshot learning with semantic output codes. In NIPS , 2009. -
- [34] D. Parikh and K. Grauman. Relative attributes. In ICCV , 2011. -
- [21] Y. Huang, Q. Liu, and D. Metaxas. Video object segmentation by hypergraph cut. In CVPR , 2009. -
Cited by
Zero-Shot Learning—A Comprehensive Evaluation of the Good, the Bad and the Ugly
Xian, Y., Lampert, C. H., Schiele, B., & Akata, Z. (2019) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2018.2857768 · Cited by 1,678
“Apart from the inductive zero-shot learning set-up where the model has no access to neither visual nor side-information of unseen classes, transductive zero-shot learning approaches [28], [44], [45], [46], [47], [48], [49] use visual or semantic information of both seen and unseen classes without having access to the…”
BackgroundFeature Generating Networks for Zero-Shot Learning
Xian, Y., Lorenz, T., Schiele, B., & Akata, Z. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00581 · Cited by 1,080
Semantic Autoencoder for Zero-Shot Learning
Kodirov, E., Xiang, T., & Gong, S. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.473 · Cited by 921
“Existing ZSL models mostly suffer from the projection domain shift problem [21].”
BackgroundSelf-citationInteractive medical image segmentation using deep learning with image-specific fine tuning
Wang, G., Li, W., Zuluaga, M. A., Pratt, R., Patel, P. A., Aertsen, M., Doel, T., David, A. L., Deprest, J., Ourselin, S., et al. (2018) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2018.2791721 · Cited by 843
Learning Deep Representations of Fine-Grained Visual Descriptions
Reed, S., Akata, Z., Lee, H., & Schiele, B. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.13 · Cited by 760
“TMVHLP [14] builds a hypergraph on a multiview embedding space learned via CCA which uses deep image features and attributes.”
BackgroundLearning a Deep Embedding Model for Zero-Shot Learning
Zhang, L., Xiang, T., & Gong, S. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.321 · Cited by 740
“Some recent methods [49, 11, 12] are tranductive in that they use all test data at once for model training, which gives them a big unfair advantage.”
MethodSelf-citationSynthesized Classifiers for Zero-Shot Learning
Changpinyo, S., Chao, W.-L., Gong, B., & Sha, F. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.575 · Cited by 703
“More recent studies show that nearest neighbor classifiers in the semantic space are very effective [10, 11, 12, 13, 30].”
BackgroundMethodResultInfluentialAn Empirical Study and Analysis of Generalized Zero-Shot Learning for Object Recognition in the Wild
Chao, W.-L., Changpinyo, S., Gong, B., & Sha, F. (2016) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-46475-6_4 · Cited by 616
“Existing research on ZSL has been almost entirely focusing on this setting [4,10,11,12,13,14,15,8,16,17,18,19,20,21,22,23,24,25,26,27,28].”
BackgroundMethodF-VAEGAN-D2: A Feature Generating Framework for Any-Shot Learning
Xian, Y., Sharma, S., Schiele, B., & Akata, Z. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.01052 · Cited by 581
“TMV [14] combines multiple semantic embeddings and performs hypergraph label propagation.”
BackgroundZheng, Z., & Yang, Y. (2021) · International Journal of Computer Vision · doi:10.1007/s11263-020-01395-y · Cited by 545
A Survey of Zero-Shot Learning: Settings, Methods, and Applications
Wang, W., Zheng, V. W., Yu, H., & Miao, C. (2019) · ACM Transactions on Intelligent Systems and Technology · doi:10.1145/3293318 · Cited by 528
A Review of Generalized Zero-Shot Learning Methods
Pourpanah, F., Abdar, M., Luo, Y., Zhou, X., Wang, R., Lim, C. P., Wang, X.-Z., & Wu, Q. M. J. (2022) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2022.3191696 · Cited by 473
“space using data samples from the seen classes without any adaptation to the unseen classes causes the projection domain shift problem [42], [63], [66].”
BackgroundMethodA Generative Model for Zero Shot Learning Using Conditional Variational Autoencoders
Mishra, A., Reddy, S. K., Mittal, A., & Murthy, H. A. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2018.00294 · Cited by 316
Attentive Region Embedding Network for Zero-Shot Learning
Xie, G.-S., Liu, L., Jin, X., Zhu, F., Zhang, Z., Qin, J., Yao, Y., & Shao, L. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00961 · Cited by 312
“An attentive region embedding network (AREN) is proposed for solving the challenging ZSL/GZSL task, which consists of two branches, i.e., the upper stream attentive region embedding (ARE) and the bottom stream (attention guided) compressed second-order embedding (ACSE).”
BackgroundInfluentialMulti-Level Semantic Feature Augmentation for One-Shot Learning
Chen, Z., Fu, Y., Zhang, Y., Jiang, Y.-G., Xue, X., & Sigal, L. (2019) · IEEE Transactions on Image Processing · doi:10.1109/tip.2019.2910052 · Cited by 254
“These additional augmented training images could be borrowed from unlabeled data [3] or other relevant categories [4], [5], [6], [7] in an unsupervised or semi-supervised fashion.”
BackgroundSelf-citationZero-Shot Visual Recognition Using Semantics-Preserving Adversarial Embedding Networks
Chen, L., Zhang, H., Xiao, J., Liu, W., & Chang, S.-F. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00115 · Cited by 309
Contrastive Embedding for Generalized Zero-Shot Learning
Han, Z., Fu, Z., Chen, S., & Yang, J. (2021) · 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr46437.2021.00240 · Cited by 291
Latent Embedding Feedback and Discriminative Features for Zero-Shot Classification
Narayan, S., Gupta, A., Khan, F. S., Snoek, C. G. M., & Shao, L. (2020) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-58542-6_29 · Cited by 260
“In recent years, the problem of object recognition under zero-shot learning (ZSL) settings has been well studied [1,9,10,16,33,34,42,45].”
Fine-Grained Generalized Zero-Shot Learning via Dense Attribute-Based Attention
Huynh, D., & Elhamifar, E. (2020) · 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr42600.2020.00454 · Cited by 250
“Unseen class bias is the direct consequence of the domain shift [41] between training and testing time, where the model overfits on classes seen at training time.”
BackgroundSemantic Projection Network for Zero- and Few-Label Semantic Segmentation
Xian, Y., Choudhury, S., He, Y., Schiele, B., & Akata, Z. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00845 · Cited by 240
“Most advances in zero-shot image classification were achieved by visualsemantic embedding models [13, 1, 57, 50, 41, 14, 55] that learn a compatibility function between the image embedding space i.”
BackgroundTransductive Unbiased Embedding for Zero-Shot Learning
Song, J., Shen, C., Yang, Y., Liu, Y., & Song, M. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00113 · Cited by 234
A Comprehensive Survey of Convolutions in Deep Learning: Applications, Challenges, and Future Trends
Younesi, A., Ansari, M., Fazli, M., Ejlali, A., Shafique, M., & Henkel, J. (2024) · IEEE Access · doi:10.1109/access.2024.3376441 · Cited by 197
“Zero-shot learning emerges as a promising area where CNNs imagine possibilities beyond the limitations of labeled data [370], [371], [372], [373], [374].”
BackgroundTransferable Contrastive Network for Generalized Zero-Shot Learning
Jiang, H., Wang, R., Shan, S., & Chen, X. (2019) · 2019 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00986 · Cited by 215
“Since the domain shift problem exists [11], the models learned on the source classes may not be suitable to the target classes, which results in overfitting the source categories in the GZSL task.”
BackgroundMethodZero-VAE-GAN: Generating Unseen Features for Generalized and Transductive Zero-Shot Learning
Gao, R., Hou, X., Qin, J., Chen, J., Liu, L., Zhu, F., Zhang, Z., & Shao, L. (2020) · IEEE Transactions on Image Processing · doi:10.1109/tip.2020.2964429 · Cited by 210
A Simple Exponential Family Framework for Zero-Shot Learning
Verma, V. K., & Rai, P. (2017) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-71246-8_48 · Cited by 199
“In another related work, [8] leverage unlabeled data in a transductive ZSL framework to handle the domain shift problem.”
BackgroundGeneralized Zero-Shot Learning with Deep Calibration Network
Liu, S., Long, M., Wang, J., & Jordan, M. I. (2018) · neural information processing systems · Cited by 267
Deep Learning for Free-Hand Sketch: A Survey
Xu, P., Hospedales, T. M., Yin, Q., Song, Y.-Z., Xiang, T., & Wang, L. (2022) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2022.3148853 · Cited by 221
“Similar to natural photo zero-shot learning/recognition [256], [257], [258], ZS-SBIR systems aim to enable query and retrieval of categories that are from unseen categories.”
BackgroundMethodSelf-citationRecent Advances in Zero-Shot Recognition: Toward Data-Efficient Understanding of Visual Content
Fu, Y., Xiang, T., Jiang, Y.-G., Xue, X., Sigal, L., & Gong, S. (2018) · IEEE Signal Processing Magazine · doi:10.1109/msp.2017.2763441 · Cited by 161
Self-citationRegion Graph Embedding Network for Zero-Shot Learning
Xie, G.-S., Liu, L., Zhu, F., Zhao, F., Zhang, Z., Yao, Y., Qin, J., & Shao, L. (2020) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-58548-8_33 · Cited by 163
“Additionally, issues with domain bias [15, 8] still exist, meaning that the learned models merely rely on the seen categories, while ignoring the available unseen attributes.”
BackgroundDiscriminative Learning of Latent Features for Zero-Shot Recognition
Li, Y., Zhang, J., Zhang, J., & Huang, K. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00779 · Cited by 173
Attribute Attention for Semantic Disambiguation in Zero-Shot Learning
Liu, Y., Guo, J., Cai, D., & He, X. (2019) · 2019 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00680 · Cited by 175
“Neither visual information nor side-information of unseen classes is avail-∗ able for inductive ZSL [49, 35, 17, 24, 16] while transductive ZSL [20, 13, 42, 47, 10, 39] has access to part of the unlabeled images.”
BackgroundMethodInfluentialZero-Shot Visual Recognition via Bidirectional Latent Embedding
Wang, Q., & Chen, K. (2017) · International Journal of Computer Vision · doi:10.1007/s11263-017-1027-5 · Cited by 167
“In addition, it has been reported that the joint use of multiple visual representations can improve the performances and the robustness of visual recognition, e.g., (Fu et al. 2015; Shao et al. 2016).”
BackgroundMethodInfluentialLearning Robust Visual-Semantic Embeddings
Tsai, Y.-H. H., Huang, L.-K., & Salakhutdinov, R. (2017) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.386 · Cited by 185
“Expand the test-time search space: Note that most of the methods [9, 41, 29, 49, 2, 50, 48, 7, 10] consider that, at test time, queries come from only test classes.”
BackgroundMethodFrom Zero-Shot Learning to Conventional Supervised Classification: Unseen Visual Data Synthesis
Long, Y., Liu, L., Shao, L., Shen, F., Ding, G., & Han, J. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.653 · Cited by 157
“Methods under different settings [40, 13, 7, 9], or using other various semantic information [36, 52, 1, 3] are not compared with.”
BackgroundLearning Deep Representations of Fine-grained Visual Descriptions
Reed, S., Akata, Z., Schiele, B., & Lee, H. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1605.05395 · Cited by 130
“TMVHLP [14] builds a hypergraph on a multiview embedding space learned via CCA which uses deep image features and attributes.”
BackgroundZero-Shot Action Recognition with Error-Correcting Output Codes
Qin, J., Liu, L., Shao, L., Shen, F., Ni, B., Chen, J., & Wang, Y. (2017) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2017.117 · Cited by 176
“semantic attributes [12, 25, 62, 14, 26] and word vectors [2, 58, 23, 15].”
BackgroundMethodZero-Shot Recognition Using Dual Visual-Semantic Mapping Paths
Li, Y., Wang, D., Hu, H., Lin, Y., & Zhuang, Y. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.553 · Cited by 149
“(2) Transductive ZSR: Due to the manifold structural information exists in unseen samples, transductive ZSR works process them in parallel and make use of the underlying manifold information to boost ZSR performance [23, 6, 15].”
BackgroundTransductive Zero-Shot Action Recognition by Word-Vector Embedding
Xu, X., Hospedales, T. M., & Gong, S. (2017) · International Journal of Computer Vision · doi:10.1007/s11263-016-0983-5 · Cited by 169
“Transductive Setting Of the four strategies, manifold regularization, self-training, and hubness correction assume access to the full set of unlabelled testing data, which is called the transductive setting (Belkin et al, 2006; Fu et al, 2015a).”
BackgroundMethodResultInfluentialSelf-citationLearning Class Prototypes via Structure Alignment for Zero-Shot Recognition
Jiang, H., Wang, R., Shan, S., & Chen, X. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-01249-6_8 · Cited by 145
Semantically Consistent Regularization for Zero-Shot Recognition
Morgado, P. M., & Vasconcelos, N. M. (2017) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2017.220 · Cited by 151
“Zero-shot learning Most current solutions to ZSL fall under two main categories: RIS and RULE.”
BackgroundZhang, J., Li, W., Ogunbona, P., & Xu, D. (2020) · ACM Computing Surveys · doi:10.1145/3291124 · Cited by 85
“(Figure used courtesy of [66])”
BackgroundInfluentialZero-Shot Learning via Class-Conditioned Deep Generative Models
Wang, W., Pu, Y., Verma, V. K., Fan, K., Zhang, Y., Chen, C., Rai, P., & Carin, L. (2018) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v32i1.11600 · Cited by 145
Compositional Learning for Human Object Interaction
Kato, K., Li, Y., & Gupta, A. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-01264-9_15 · Cited by 144
“Fu et al. [16] used hyper-graph label propagation to fuse information from multiple semantic representations.”
BackgroundSemi-supervised Vocabulary-Informed Learning
Fu, Y., & Sigal, L. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.576 · Cited by 128
“ZSL was first explored with use of attributebased semantic representations [11, 15, 17, 18, 24, 32].”
MethodSelf-citationLow-Rank Embedded Ensemble Semantic Dictionary for Zero-Shot Learning
Ding, Z., Shao, M., & Fu, Y. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.636 · Cited by 126
“The learned common space is either interpretable [36] or latent [9].”
BackgroundSelf-citationContext-aware Feature Generation For Zero-shot Semantic Segmentation
Gu, Z., Zhou, S., Niu, L., Zhao, Z., & Zhang, L. (2020) · Proceedings of the 28th ACM International Conference on Multimedia · doi:10.1145/3394171.3413593 · Cited by 112
“Many ZSL methods [1, 8, 9, 11, 27, 32, 33, 37, 46] attempted to learn a mapping between feature space and semantic embedding space.”
MethodMatrix Tri-Factorization with Manifold Regularizations for Zero-Shot Learning
Xu, X., Shen, F., Yang, Y., Zhang, D., Shen, H. T., & Song, J. (2017) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2017.217 · Cited by 139
Heterogeneous Knowledge Transfer in Video Emotion Recognition, Attribution and Summarization
Xu, B., Fu, Y., Jiang, Y.-G., Li, B., & Sigal, L. (2016) · IEEE Transactions on Affective Computing · doi:10.1109/taffc.2016.2622690 · Cited by 135
“In addition, we use cosine similarity as the metric rather than the euclidean distance since the semantic word vectors are intrinsically directional and cosine similarity is a better metric used in [40], [73].”
MethodSelf-citationSemantics-Guided Contrastive Network for Zero-Shot Object Detection
Yan, C., Chang, X., Luo, M., Liu, H., Zhang, X., & Zheng, Q. (2022) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2021.3140070 · Cited by 137
“As a result, the learned model can easily bias towards the seen categories, leading to limited transferable ability of the model [16], [17], [18].”
BackgroundA review on multi-view learning
Yu, Z., Dong, Z., Yu, C., Yang, K., Fan, Z., & Chen, C. L. P. (2024) · Frontiers of Computer Science · doi:10.1007/s11704-024-40004-w · Cited by 147
Multi-Task Zero-Shot Action Recognition with Prioritised Data Augmentation
Xu, X., Hospedales, T. M., & Gong, S. (2016) · Lecture notes in computer science · doi:10.1007/978-3-319-46475-6_22 · Cited by 131
“Commonly used visual-semantic mappings include linear [12] and non-linear regression [6,9,11], classification [5,21], and bilinear ranking [7].”
BackgroundMethodInfluentialSelf-citationModeling Inter and Intra-Class Relations in the Triplet Loss for Zero-Shot Learning
Cacheux, Y. L., Borgne, H. L., & Crucianu, M. (2019) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.01043 · Cited by 129
“Transductive ZSL methods [13, 17, 30, 34] assume that unlabeled samples from unseen classes are available during training.”
Backgroundp-Laplacian Regularized Sparse Coding for Human Activity Recognition
Liu, W., Zha, Z.-J., Wang, Y., Lü, K., & Tao, D. (2016) · IEEE Transactions on Industrial Electronics · doi:10.1109/tie.2016.2552147 · Cited by 116
“[71], [74], 2) the efficient utilization of multiview features [66] [73], and 3) the deep architecture are valuable to improve the performance.”
BackgroundLearning Discriminative Latent Attributes for Zero-Shot Classification
Jiang, H., Wang, R., Shan, S., Yang, Y., & Chen, X. (2017) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.453 · Cited by 120
“Note that ’*’ indicates that the approaches use different features from ours, where [11, 15] use the OverFeat [34] features and [2, 5, 6] use the GoogleNet [38] features.”
BackgroundMethodInfluentialStacked Semantic-Guided Attention Model for Fine-Grained Zero-Shot LearningPreprint
yu, Y., Ji, Z., Fu, Y., Guo, J., Pang, Y., & Zhang, Z. M. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1805.08113 · Cited by 124
“Targeting on extending conventional classification models to unseen classes, zero-shot learning (ZSL) [3, 10, 9, 12, 34, 35, 36] has attracted a lot of interests in the machine learning and computer vision communities.”
BackgroundSelf-citationBaby steps towards few-shot learning with multiple semantics
Schwartz, E., Karlinsky, L., Feris, R., Giryes, R., & Bronstein, A. M. (2022) · Pattern Recognition Letters · doi:10.1016/j.patrec.2022.06.012 · Cited by 121
Zero-Shot Learning Using Synthesised Unseen Visual Data with Diffusion Regularisation
Long, Y., Liu, L., Shen, F., Shao, L., & Li, X. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2017.2762295 · Cited by 115
“However, rather than concerning the ‘domainshift problem’ for the recognition task in [23], instead, we hope the model can effectively capture the semantic-visual correlation so that the synthesised visual data can preserve the intrinsic structure as close as the real data.”
BackgroundMethodResultInfluentialGeneralized Multi-View Embedding for Visual Recognition and Cross-Modal Retrieval
Cao, G., Iosifidis, A., Chen, K., & Gabbouj, M. (2017) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2017.2742705 · Cited by 108
“5) Zero-Shot Recognition on the AwA Dataset: We follow a similar experiment pipeline as in [57], and the comparative results show the performance of the proposed multi-view embedding methods.”
MethodInfluentialZero-Shot Scene Classification for High Spatial Resolution Remote Sensing Images
Li, A., Lu, Z., Wang, L., Xiang, T., & Wen, J.-R. (2017) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2017.2689071 · Cited by 101
Self-citationMulti-task Curriculum Transfer Deep Learning of Clothing Attributes
Dong, Q., Gong, S., & Zhu, X. (2017) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv.2017.64 · Cited by 105
“zero-shot learning [35, 19], face analysis [34], pedestrian description [10, 20], person reidentification [36], visual search [31, 49, 16].”
Self-citationRelational Knowledge Transfer for Zero-Shot Learning
Wang, D., Li, Y., Lin, Y., & Zhuang, Y. (2016) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v30i1.10195 · Cited by 102
“To further improve the ZSL performance, (Fu et al. 2015a) combines multiple semantic spaces and propagated label predictions on multiple graphs.”
BackgroundMethodInfluentialJoint Semantic and Latent Attribute Modelling for Cross-Class Transfer Learning
Peng, P., Tian, Y., Xiang, T., Wang, Y., Pontil, M., & Huang, T. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2017.2723882 · Cited by 102
“For the ZSL problem, each unseen class is given an attribute “prototype” [11] and”
MethodSelf-citationTransductive Zero-Shot Learning With a Self-Training Dictionary Approach
Yu, Y., Ji, Z., Li, X., Guo, J., Zhang, Z. M., Ling, H., & Wu, F. Q. (2018) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2017.2751741 · Cited by 96
Zero-Shot Learning on Semantic Class Prototype Graph
Fu, Z., Xiang, T., Kodirov, E., & Gong, S. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2017.2737007 · Cited by 98
“In fact, as shown in [24], the projected unseen class data points are often biased towards some seen class prototypes and far away from the unseen class prototypes they belong, resulting in poor recognition performance.”
BackgroundMethodInfluentialSelf-citationAttribute-Guided Network for Cross-Modal Zero-Shot Hashing
Ji, Z., Sun, Y., Yu, Y., Pang, Y., & Han, J. (2020) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2019.2904991 · Cited by 94
“In addition, under the transductive setting [25], [26], Lai et al.”
BackgroundZero-Shot Recognition via Structured Prediction
Zhang, Z., & Saligrama, V. (2016) · Lecture notes in computer science · doi:10.1007/978-3-319-46478-7_33 · Cited by 97
“Compared against related methods such as [14, 15] which both benefit from exploring data structures like ours, our method significantly outperforms these methods by 11.”
BackgroundMethodInfluentialTransductive Learning for Zero-Shot Object Detection
Rahman, S., Khan, S., & Barnes, N. (2019) · 2019 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00618 · Cited by 84
“This problem is called the ‘domain-shift problem’ and poses a significant challenge to generalization of a ZSL approach [5].”
BackgroundTransZero++: Cross Attribute-Guided Transformer for Zero-Shot Learning
Chen, S., Hong, Z., Hou, W., Xie, G.-S., Song, Y., Zhao, J. C., You, X., Yan, S., & Shao, L. (2022) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2022.3229526 · Cited by 133
“Motivated by this challenge, zero-shot learning (ZSL) was proposed to recognize new classes by exploiting the intrinsic semantic relatedness during learning [1], [2], [3], [4], [5], [6].”
Disentangle, Align and Fuse for Multimodal and Semi-Supervised Image Segmentation
Chartsias, A., Papanastasiou, G., Wang, C., Semple, S., Newby, D. E., Dharmakumar, R., & Tsaftaris, S. A. (2020) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2020.3036584 · Cited by 96
Multi-View Representation Learning With Deep Gaussian Processes
Sun, S., Dong, W., & Liu, Q. (2020) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2020.3001433 · Cited by 86
Zero and Few Shot Learning With Semantic Feature Synthesis and Competitive Learning
Guan, J., Lu, Z., Xiang, T., Li, A., Zhao, A., & Wen, J.-R. (2020) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2020.2965534 · Cited by 83
“(3) Transductive ZSL: the unlabelled unseen class samples are used for model training [15], [16], [17], [18], [19].”
BackgroundSelf-citationZero-shot Text Classification via Reinforced Self-training
Ye, Z., Geng, Y., Chen, J., Chen, J., Xu, X., Zheng, S., Wang, F., Zhang, J., & Chen, H. (2020) · Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/2020.acl-main.272 · Cited by 76
“(2) due to the severe domain shift (Fu et al., 2015), traditional self-training method may not provide reliable selection.”
BackgroundInfluentialProgressive Ensemble Networks for Zero-Shot Recognition
Ye, M., & Guo, Y. (2019) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2019.01200 · Cited by 86
“…methods identify visual-semantic mappings only on the labeled seen class data, which poses a fundamental domain shift problem due to the appearance variations of visual attributes across seen and unseen classes, and has negative impact on cross-class generalization (i.e., ZSL performance) [11, 18].”
BackgroundMeasuring and Improving User Experience Through Artificial Intelligence-Aided Design
Yang, B., Wei, L., & Pu, Z. (2020) · Frontiers in Psychology · doi:10.3389/fpsyg.2020.595374 · Cited by 86
“Based on this, the zero-shot learning (Fu et al., 2015) technique can be used to synthesize the query-generated retrieval model (Huang et al., 2016).”
MethodOntoZSL: Ontology-enhanced Zero-shot Learning
Geng, Y., Chen, J., Chen, Z., Pan, J. Z., Ye, Z., Yuan, Z., Jia, Y., & Chen, H. (2021) · International World Wide Web Conference (WWW) · doi:10.1145/3442381.3450042 · Cited by 100
“As shown in the left of Figure 1, in general, textual descriptions may not provide distinguishing characteristics necessary for ZSL due to the noisy words in text [33]; attribute descriptions focus on the “local” characteristics of objects and easily suffer from the domain shift problem [12]; while label taxonomy mere…”
BackgroundVisual Structure Constraint for Transductive Zero-Shot Learning in the Wild
Wan, Z., Chen, D., & Liao, J. (2021) · International Journal of Computer Vision · doi:10.1007/s11263-021-01451-1 · Cited by 78
“This problem is first witnessed by Fu et al. (2015) and alleviated by a new transductive zero-shot learning setting.”
BackgroundMethodInfluentialUnsupervised Heterogeneous Domain Adaptation via Shared Fuzzy Equivalence Relations
Liu, F., Lü, J., & Zhang, G. (2018) · IEEE Transactions on Fuzzy Systems · doi:10.1109/tfuzz.2018.2836364 · Cited by 74
“…help classify English documents [6]; building recognition models capable of recognising novel visual categories without labelled training samples [7]; detecting a user’s current location based on previously collected WiFi data [3] and leveraging the large number of labeled simple actions to…”
BackgroundZero-Shot Learning via Latent Space Encoding
Yu, Y., Ji, Z., Guo, J., & Zhang, Z. M. (2018) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2018.2850750 · Cited by 78
Self-Supervised Domain-Aware Generative Network for Generalized Zero-Shot Learning
Wu, J., Zhang, T., Zha, Z.-J., Luo, J., Zhang, Y., & Wu, F. (2020) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr42600.2020.01278 · Cited by 80
“In this paper, we apply GZSL in a transductive way, assuming that unlabeled target data is provided in training [12].”
BackgroundMethodZero-Shot Learning of SAR Target Feature Space With Deep Generative Neural Networks
Song, Q., & Xu, F. (2017) · IEEE Geoscience and Remote Sensing Letters · doi:10.1109/lgrs.2017.2758900 · Cited by 72
Zero-Shot Learning on 3D Point Cloud Objects and Beyond
Cheraghian, A., Rahman, S., Chowdhury, T. F., Campbell, D., & Petersson, L. (2022) · International Journal of Computer Vision · doi:10.1007/s11263-022-01650-4 · Cited by 63
“[15] proposed a multi-view transductive setting to address projection shift and to exploit various semantic representations of the visual feature.”
Hybrid-Attention Based Decoupled Metric Learning for Zero-Shot Image Retrieval
Chen, B., & Deng, W. (2019) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2019.00286 · Cited by 75
“Relation to other attention mechanism: Many other attention works, such as spatial-attention [40, 8], channelattention [4, 12] and semantic-attention [42, 30], employ the task-specific attention modules to mine the most informative signals.”
BackgroundFuzzy Rule-Based Domain Adaptation in Homogeneous and Heterogeneous Spaces
Zuo, H., Lü, J., Zhang, G., & Witold Pedrycz, C. (2018) · IEEE Transactions on Fuzzy Systems · doi:10.1109/tfuzz.2018.2853720 · Cited by 66
“Some examples include: using already-categorized French documents to help classify English documents [2]; building recognition models capable of identifying novel visual categories [3]; and detecting a user’s current location given previously collected WiFi data [4].”
Gao, Y., Ji, S., Han, X., & Dai, Q. (2024) · Engineering · doi:10.1016/j.eng.2024.04.017 · Cited by 63
Transductive Zero-Shot Learning for 3D Point Cloud Classification
Cheraghian, A., Rahman, S., Campbell, D., & Petersson, L. (2020) · 2020 IEEE Winter Conference on Applications of Computer Vision (WACV) · doi:10.1109/wacv45572.2020.9093545 · Cited by 61
“This gives rise to the problem of projection domain shift [14].”
InfluentialZero-Shot Learning via Attribute Regression and Class Prototype Rectification
Luo, C., Li, Z., Huang, K., Feng, J., & Wang, M. (2017) · IEEE Transactions on Image Processing · doi:10.1109/tip.2017.2745109 · Cited by 66
Zero Shot Learning via Low-rank Embedded Semantic AutoEncoder
Liu, Y., Gao, Q., Li, J., Han, J., & Shao, L. (2018) · International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2018/345 · Cited by 65
“This can lead to a projection domain shift problem [Fu et al. , 2015a].”
BackgroundMethodClassifier and Exemplar Synthesis for Zero-Shot Learning
Changpinyo, S., Chao, W., Gong, B., & Sha, F. (2019) · International Journal of Computer Vision · doi:10.1007/s11263-019-01193-1 · Cited by 64
“heir choices of common spaces or compatibility functions. The linear or bilinear compatibility functions are extensively used [2,3,17, 61,40]. Some propose to use canonical correlation analysis (CCA) [18,19,42]. Nonlinear methods are scarce but have also been explored such as dictionary learning and sparse coding [33,…”
BackgroundGenerative Zero-Shot Learning via Low-Rank Embedded Semantic Dictionary
Ding, Z., Shao, M., & Fu, Y. (2018) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2018.2867870 · Cited by 61
“In general, there are three lines of approaches as follows: (1) direct mapping [19], [20] including semantic coding methods [2], [11], [21]; (2) parameter mapping [1], [22]; and (3) common space learning [3], [23].”
BackgroundSelf-citationGNDAN: Graph Navigated Dual Attention Network for Zero-Shot Learning
Chen, S., Hong, Z., Xie, G.-S., Peng, Q., You, X., Ding, W., & Shao, L. (2022) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2022.3155602 · Cited by 71
“Limited by the issue of domain bias [43], [44], these methods typically perform poorly in the GZSL setting.”
MethodRecognizing an Action Using Its Name: A Knowledge-Based Approach
Gan, C., Yang, Y., Zhu, L., Zhao, D., & Zhuang, Y. (2016) · International Journal of Computer Vision · doi:10.1007/s11263-016-0893-6 · Cited by 65
“Besides, severalworks proposed attributes as themid-level representation and also applied it to zero-shot action recognition (Laptev 2005; Rohrbach et al. 2012; Fu et al. 2014, 2015), and fewshot (Rohrbach et al. 2013) action recognition.”
BackgroundGenerative Multi-Label Zero-Shot Learning
Gupta, A., Narayan, S., Khan, S. H., Khan, F. S., Shao, L., & van de Weijer, J. (2023) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2023.3295772 · Cited by 61
Tavakoli, M., Mehdizadeh, A., Aghayan, A., Pourreza, R., Ellis, T., & Dehmeshki, J. (2021) · IEEE Access · doi:10.1109/access.2021.3074458 · Cited by 58
“Our future plan is to investigate the combination deep neural networks and N-shot learning [92], [93] with the methods used in this study to see if results can be further improved.”
MethodCross-Lingual Pre-Training Based Transfer for Zero-Shot Neural Machine Translation
Ji, B., Zhang, Z., Duan, X., Zhang, M., Chen, B., & Luo, W. (2020) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v34i01.5341 · Cited by 56
A review on multimodal zero‐shot learning
Cao, W., Wu, Y., Sun, Y., Zhang, H., Ren, J., Gu, D., & Wang, X. (2023) · Wiley Interdisciplinary Reviews Data Mining and Knowledge Discovery · doi:10.1002/widm.1488 · Cited by 68
“Fu et al. (2015) investigated the reasons for prototype sparsity and projection domain shift problems encountered by existing ZSL algorithms, that is, the semantic representation of categories is single and there is no correlation between some seen and unseen classes, and proposed solutions…”
BackgroundInfluentialOSLNet: Deep Small-Sample Classification With an Orthogonal Softmax Layer
Li, X., Chang, D., Ma, Z., Tan, Z., Xue, J., Cao, J., Yu, J., & Guo, J. H. (2020) · IEEE Transactions on Image Processing · doi:10.1109/tip.2020.2990277 · Cited by 59
Video Emotion Recognition with Transferred Deep Feature Encodings
Xu, B., Fu, Y., Jiang, Y.-G., Li, B., & Sigal, L. (2016) · Proceedings of the 2016 ACM on International Conference on Multimedia Retrieval · doi:10.1145/2911996.2912006 · Cited by 53
“First, due to close interaction between cognitive processes and emotional appraisals [29, 13, 14], human emotions are rich and complex.”
BackgroundSelf-citationAttribute-Based Synthetic Network (ABS-Net): Learning more from pseudo feature representations
Lu, J., Li, J., Yan, Z., Mei, F., & Zhang, C. (2018) · Pattern Recognition · doi:10.1016/j.patcog.2018.03.006 · Cited by 53
A Large-Scale Attribute Dataset for Zero-Shot Learning
Zhao, B., Fu, Y., Liang, R. S., Wu, J., Wang, Y., & Wang, Y. (2019) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2019.00053 · Cited by 65
“Thus the generalized attributes may not fit visual representations well, but they are complementary to user-defined attributes [13].”
BackgroundSelf-citationDeep Unbiased Embedding Transfer for Zero-Shot Learning
Jia, Z., Zhang, Z., Wang, L., Shan, C., & Tan, T. N. (2019) · IEEE Transactions on Image Processing · doi:10.1109/tip.2019.2947780 · Cited by 56
“Nevertheless, these methods suffer from the projection domain shift problem [19].”
Background
Metadata
- Authors
- Fu, Y., Hospedales, T. M., Xiang, T., & Gong, S.
- Affiliations
- Walt Disney (United States) (US)
- Queen Mary University of London (GB)
- Venue
- IEEE Transactions on Pattern Analysis and Machine Intelligence
- Published
- 2015-11-01
- Volume
- 37
- Issue
- 11
- Pages
- 2332-2345
- ISSN
- 0162-8828 (ISSN-L)
- 2160-9292
- OpenAlex
- W2141350700
- Cited by
- 514
- References
- 74
- Citations by year
- 2026: 17 · 2025: 32 · 2024: 32 · 2023: 52 · 2022: 60 · 2021: 75 · 2020: 62 · 2019: 57 · 2018: 64 · 2017: 41 · 2016: 21 · 2015: 1
- FWCI
- 41.51
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- IEEE Transactions on Pattern Analysis and Machine Intelligence (publishedVersion)
- PubMed (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- Queen Mary Research Online (Queen Mary University of London) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Image and Video Retrieval Techniques
- Keywords
- zero-shot learning, transductive learning, multi-view embedding, semantic representation, transfer learning, manifold structure
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.