Pith. sign in

REVIEW 4 major objections 5 minor 53 references

Visual and Semantic Prototypes-Jointly Guided CNN for Generalized Zero-shot Learning

T0 review · 4 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Generalized zero-shot learning is decomposed into a two-stage pipeline—reject what is not a known class, then identify the rejected image by its attributes—and the authors show this split solves the class overfitting problem that plagues…

desk verdict A clearly written domain-separating G-ZSL paper with a real joint-training ablation, but the 'effectively solves' claim overreaches and the entropy gate is under-analyzed. read the letter →

arxiv 1908.03983 v2 pith:FTC7TWVM submitted 2019-08-12 cs.LG cs.CVstat.ML

Chuanxing Geng , Lue Tao , Songcan Chen This is my paper · ORCID
classification cs.LGcs.CVstat.ML
keywords generalizedzero-shotlearningopensetrecognitionconvolutionalprototypevisual-semanticembeddingclassoverfittingentropy-basedrejectionattribute-basedclassificationdomainseparation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that generalized zero-shot learning (G-ZSL) — recognizing both classes seen in training and classes never seen, using shared attribute descriptions — is best solved by splitting it into two tasks rather than mixing seen and unseen classes in one classifier. A first branch, built on convolutional prototype learning, recognizes seen classes from visual features and rejects anything that does not look like a seen class; the rejected images are then handed to a second branch that matches them to semantic attribute prototypes of unseen classes. The authors claim this decomposition removes the class overfitting problem, where unseen images get forced into seen classes, and they support it with a network called VSG-CNN that trains both branches jointly. On four standard benchmarks, VSG-CNN reaches the best harmonic-mean accuracy on CUB, AWA2, and aPY, and is competitive on SUN.

What carries the argument

The central object is VSG-CNN, a convolutional network with two branches built on a shared frozen feature extractor. A visual-prototype branch (a convolutional prototype network) learns a set of per-class visual prototypes and assigns probabilities by distance-based cross-entropy; a semantic branch maps the same shared features to predefined class attribute vectors. The load-bearing mechanism at test time is an entropy gate: the probability distribution of the test image over seen-class visual prototypes is computed, and its entropy decides the domain—low entropy means 'seen', high entropy means 'unseen'. The threshold is set on a validation split. This gate is what separates the two tasks and is claimed to eliminate the class overfitting problem.

What would settle it

On a held-out validation set, compute the entropy values for seen and unseen images under the learned visual prototypes and check whether a single threshold separates them. If the entropy distributions overlap substantially (e.g., more than a few percent of unseen images fall below the best seen/unseen threshold), then no threshold can prevent unseen images from being absorbed into seen classes, and the claimed solution to class overfitting fails on that data.

Watch

Extended reading notes

Core claim

The paper's central claim is that the class overfitting problem in G-ZSL can be effectively solved by decomposing the task into open set recognition (OSR) followed by zero-shot learning (ZSL). Seen classes are recognized directly in visual space via learned visual prototypes; unseen classes first fail a visual-prototype acceptance test and are then classified in semantic space against predefined attribute prototypes. This separation, implemented in the jointly trained VSG-CNN, means unseen images are never forced into seen classes, and seen images retain the richer discriminability of visual features. The same architecture is repurposed for a new task the paper calls generalized open set recognition (G-OSR), where rejected unknowns are not merely discarded but given a rough attribute description derived from known-class semantic information. On CUB, AWA2, SUN, and aPY, the method reports state-of-the-art or competitive harmonic-mean accuracy, with notably smaller gaps between seen and unseen accuracy than visual-semantic embedding baselines.

Load-bearing premise

The method assumes that the entropy of a test image's distribution over seen-class visual prototypes cleanly separates seen from unseen instances; if an unseen image happens to be visually close to a seen prototype and produces a low-entropy score, it will be misclassified as seen and the class overfitting problem reappears.

Editorial extensions

If this is right

  • Generative models for synthesizing unseen-class instances become unnecessary, sidestepping the blurriness, mode collapse, and instability that limit VAE and GAN approaches.
  • Seen classes are classified in visual space, preserving subclass discriminability that semantic embeddings blur, as the paper's t-SNE visualization illustrates.
  • Open set recognition gains an output beyond rejection: rejected unknowns receive a rough attribute description assembled from known-class semantic knowledge.
  • Joint end-to-end training of the two branches beats training them separately, as the Baseline comparison shows, so the two tasks complement each other.
  • The smaller seen-unseen accuracy gaps on most datasets indicate the class overfitting problem is mitigated, not just accuracy improved.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same two-branch decomposition should transfer to few-shot learning, where base classes are handled by the visual branch and novel classes by the semantic branch; the entropy gate would need retraining for the new class set.
  • Because the semantic branch is a simple regression to attribute vectors, swapping it for a more powerful alignment (e.g., a semantic autoencoder) could push harmonic mean higher; the paper explicitly allows this substitution.
  • A natural evaluation for G-OSR would measure how useful the output attribute vectors are to a human identifying the unknown class, rather than only rejection accuracy.
  • On datasets where seen and unseen classes are visually very similar, the entropy distributions may overlap; testing on such a dataset would show whether the claimed CO solution degrades gracefully or collapses.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper proposes VSG-CNN, a unified end-to-end framework that decomposes generalized zero-shot learning (G-ZSL) into an open set recognition (OSR) task and a zero-shot learning (ZSL) task, and also introduces a new generalized open set recognition (G-OSR) setting. The network has a shared layer, a visual prototype branch (based on convolutional prototype learning, CPL), and a semantic prototype branch for attribute-based classification. At test time, the entropy of the distance-based softmax over seen-class visual prototypes (Eq. 1) is compared with a threshold to decide whether an instance is seen or unseen; seen instances are classified by nearest visual prototype, unseen instances by nearest semantic prototype. Experiments on CUB, AWA2, SUN, and aPY report harmonic-mean accuracies for G-ZSL compared with 20 non-generative and 8 generative methods, and a G-OSR comparison against CPL only. The paper claims that this decomposition effectively solves the class overfitting (CO) problem and that VSG-CNN is the first method to exploit semantic information of known classes in OSR.

Significance. If the central claim were fully supported, the decomposition of G-ZSL into OSR plus ZSL with joint training would be a notable alternative to generative-data-augmentation approaches, and the introduction of G-OSR with semantic descriptions of unknowns would open a useful new direction. The method is clearly specified, uses standard data splits, and is compared with a broad set of recent G-ZSL methods, which are strengths. However, the load-bearing component—the entropy threshold that separates seen from unseen instances—is not theoretically grounded or quantitatively validated, and the G-OSR evaluation rests on a single baseline. The mixed results on SUN and aPY further weaken the general claim that the CO problem is effectively solved. The paper ships no code and reports no error bars, limiting the reproducibility of the numerical claims.

major comments (4)
  1. [§III.C.2, Eq. (1)] The entropy gate is the only mechanism that separates seen and unseen instances, yet the paper provides no quantitative evidence that the entropy of the distance-based softmax in Eq. (1) actually separates the two domains. Fig. 1 shows t-SNE of prototypes, not the entropy distributions of test instances, and the G-ZSL-val split described in §IV.A.2 and Fig. 4 contains only seen classes, so the threshold δ_g cannot be validated against any unseen instance. Consequently, the central claim that the decomposition 'effectively solves the CO problem' is not established.
  2. [§IV.A.2] The threshold grid for δ_g and δ_o is [0, 0.02] with step 0.000002, which is several orders of magnitude smaller than the natural entropy scale of Eq. (1) (e.g., log 150 ≈ 5.01 for CUB). No sensitivity analysis with respect to the threshold is reported, so it is unclear whether the final results are stable or artifacts of a very narrow, dataset-specific operating point.
  3. [Table 4] The G-OSR evaluation compares VSG-CNN with only CPL, and the differences are marginal: 70.5→73.3 on CUB, 84.5→84.3 on AWA2 (a slight decrease), 40.3→42.3 on SUN, and 73.6→74.8 on aPY. With no error bars, significance tests, or additional OSR baselines, the claim that introducing semantic information from known classes improves OSR is not demonstrated.
  4. [Tables II and III] On SUN, VSG-CNN is substantially worse than the best competitor (H = 30.9 vs 41.0 in Table II and vs 40.6 in Table III), and on aPY it is comparable to or worse than several visual-semantic embedding methods (e.g., H = 34.0 vs 36.6). The paper acknowledges these failures but does not explain why the decomposition does not help there, and the abstract and conclusion nevertheless assert that the CO problem is effectively solved in general.
minor comments (5)
  1. [Table II caption] The caption says 'COMPARING CRL' but this appears to be a typo; it should likely read 'COMPARING CPL' or 'COMPARING VSG-CNN'.
  2. [Fig. 5] There are typos in the attribute descriptions: 'longnech' and 'nech' should be 'longneck' and 'neck'.
  3. [§IV.A.2] The paper states that code will be published later but does not report learning rate, batch size, number of epochs, or other training details; these are needed for reproducibility, especially because the reported gains on some datasets are small.
  4. [Table II, Baseline] The 'Baseline (ours)' row is described only as a 'non-co-learning strategy'; the exact training procedure and architecture differences from VSG-CNN should be specified for the comparison to be meaningful.
  5. [§III.A] The definition of G-OSR in item 4) is informal; a formal formulation of the task and its evaluation protocol would help clarify the claimed novelty.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the entropy-gated decomposition is an empirical method with tuned hyperparameters, not a derivation that reduces to its inputs.

full rationale

The paper's central claim is empirical rather than circular. The training objective in Eq. (4) combines a convolutional prototype loss for the visual branch and a visual-semantic mapping loss for the semantic branch, with the visual-prototype distribution defined by the standard distance-based softmax in Eq. (1). The seen-versus-unseen decision is made by comparing the entropy of that distribution to thresholds delta_g and delta_o, which are selected by grid search on the G-ZSL-val validation split described in Section IV.A.2. This is legitimate hyperparameter tuning, not a fitted quantity being renamed as a prediction, and the reported H values are external benchmark results rather than consequences of the definitions. The only self-citation, reference [6] in the related-work enumeration of OSR methods, is not load-bearing for any claimed result, and no uniqueness theorem or derived quantity is imported from the authors' prior work. Concerns such as the sensitivity of the entropy threshold and the absence of a theoretical guarantee that low-entropy unseen instances will be rejected are correctness and validation risks, not circularity.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

No new physical or conceptual entities are introduced; G-OSR is a task formulation, not an entity. The framework relies on existing visual and semantic prototypes.

free parameters (5)
  • δg = not reported; grid-searched in [0, 0.02]
    Domain threshold for seen/unseen decision in G-ZSL; selected on G-ZSL-val.
  • δo = not reported; same range
    Domain threshold for known/unknown decision in G-OSR; selected on G-ZSL-val.
  • λ = not reported; grid-searched in {1e-3, 1e-2, 1e-1, 1, 10}
    Regularization weight for prototype loss in Eq. 3; tuned on validation set.
  • γ = 1
    Distance scale hyperparameter in Eq. 1; fixed by hand.
  • t = not reported; grid-searched in {32,...,717}
    Dimension of visual prototypes; tuned on validation set.
assumptions (5)
  • domain assumption Semantic attribute vectors are available for both seen and unseen classes in G-ZSL.
    Standard ZSL setting; used throughout Section III.
  • ad hoc to paper The entropy of the distance-based probability distribution separates seen from unseen instances.
    No formal proof; the method's domain detector rests on this empirical premise.
  • domain assumption Frozen early layers of ImageNet-pretrained ResNet-101 provide useful visual features.
    The shared feature extractor is pretrained and mostly frozen; if these features are not discriminative, both branches suffer.
  • domain assumption The CPL loss functions (DCE and PL) are taken as valid for learning visual prototypes.
    Paper inherits Eq. 1-3 from [5] without modification.
  • domain assumption Standard data splits from [45] are used.
    The paper follows the widely used splits; it does not introduce new data.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Visual and Semantic Prototypes-Jointly Guided CNN for Generalized Zero-shot Learning." pith.science (2026). https://pith.science/paper/FTC7TWVM

@misc{pith2026190803983,
  author       = {Pith},
  title        = {Pith review of: Visual and Semantic Prototypes-Jointly Guided CNN for Generalized Zero-shot Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FTC7TWVM}},
  note         = {Machine review of arXiv:1908.03983}
}
read the original abstract

In the process of exploring the world, the curiosity constantly drives humans to cognize new things. Supposing you are a zoologist, for a presented animal image, you can recognize it immediately if you know its class. Otherwise, you would more likely attempt to cognize it by exploiting the side-information (e.g., semantic information, etc.) you have accumulated. Inspired by this, this paper decomposes the generalized zero-shot learning (G-ZSL) task into an open set recognition (OSR) task and a zero-shot learning (ZSL) task, where OSR recognizes seen classes (if we have seen (or known) them) and rejects unseen classes (if we have never seen (or known) them before), while ZSL identifies the unseen classes rejected by the former. Simultaneously, without violating OSR's assumptions (only known class knowledge is available in training), we also first attempt to explore a new generalized open set recognition (G-OSR) by introducing the accumulated side-information from known classes to OSR. For G-ZSL, such a decomposition effectively solves the class overfitting problem with easily misclassifying unseen classes as seen classes. The problem is ubiquitous in most existing G-ZSL methods. On the other hand, for G-OSR, introducing such semantic information of known classes not only improves the recognition performance but also endows OSR with the cognitive ability of unknown classes. Specifically, a visual and semantic prototypes-jointly guided convolutional neural network (VSG-CNN) is proposed to fulfill these two tasks (G-ZSL and G-OSR) in a unified end-to-end learning framework. Extensive experiments on benchmark datasets demonstrate the advantages of our learning framework.

Figures

Figures reproduced from arXiv: 1908.03983 by the authors.

Figure 1
Figure 1. Visualization of the seen-class prototypes on validation set of aPY dataset in semantic space (a) and visual space (b) by t-SNE, where the prototypes in [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Overview of the VSG-CNN framework. The Shared Layer and the Visual Layer constitute the convolutional prototype subnetwork as one branch, [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. (a) and (b) respectively describe the prediction processes in G-ZSL and G-OSR. Different from G-ZSL, nothing about the unknown classes is available [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: G-ZSL-val split. The data is organized across classes and instances. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Cognizing unknown class. VSG-CNN utilizes the semantic/attribute information from seen classes to provide a rough semantic/attribute description [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 48 canonical work pages

  1. [1]

    Zero-shot learning through cross-modal transfer,

    R. Socher, M. Ganjoo, C. D. Manning, and A. Ng, “Zero-shot learning through cross-modal transfer,” in Neural Information Processing Sys- tems, pp. 935–943, 2013

    work page 2013

  2. [2]

    An empirical study and analysis of generalized zero-shot learning for object recognition in the wild,

    W.-L. Chao, S. Changpinyo, B. Gong, and F. Sha, “An empirical study and analysis of generalized zero-shot learning for object recognition in the wild,” in European Conference on Computer Vision , pp. 52–68, 2016

    work page 2016

  3. [3]

    Toward open set recognition,

    W. J. Scheirer, A. de Rezende Rocha, A. Sapkota, and T. E. Boult, “Toward open set recognition,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 35, no. 7, pp. 1757–1772, 2013

    2013

  4. [4]

    Probability models for open set recognition,

    W. J. Scheirer, L. P. Jain, and T. E. Boult, “Probability models for open set recognition,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 36, no. 11, pp. 2317–2324, 2014

    2014

  5. [5]

    Robust classification with convolutional prototype learning,

    H.-M. Yang, X.-Y . Zhang, F. Yin, and C.-L. Liu, “Robust classification with convolutional prototype learning,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pp. 3474– 3482, 2018

    work page 2018

  6. [6]

    Recent Advances in Open Set Recognition: A Survey

    C. Geng, S.-j. Huang, and S. Chen, “Recent advances in open set recognition: A survey,” arXiv:1811.08581, 2018

    work page Pith review arXiv 2018

  7. [7]

    Reducing network agnos- tophobia,

    A. R. Dhamija, M. G ¨unther, and T. Boult, “Reducing network agnos- tophobia,” in Neural Information Processing Systems , pp. 9157–9168, 2018

    work page 2018

  8. [8]

    C2ae: Class conditioned auto-encoder for open-set recognition,

    P. Oza and V . M. Patel, “C2ae: Class conditioned auto-encoder for open-set recognition,”IEEE Conference on Computer Vision and Pattern Recognition, in press , 2019

    work page 2019

Show all 53 references
  1. [9]

    Triple verification network for generalized zero-shot learning,

    H. Zhang, Y . Long, Y . Guan, and L. Shao, “Triple verification network for generalized zero-shot learning,” IEEE Transactions on Image Pro- cessing, vol. 28, no. 1, pp. 506–517, 2019

    2019

  2. [10]

    A unified approach for conventional zero-shot, generalized zero-shot, and few-shot learning,

    S. Rahman, S. Khan, and F. Porikli, “A unified approach for conventional zero-shot, generalized zero-shot, and few-shot learning,” IEEE Transac- tions on Image Processing , vol. 27, no. 11, pp. 5652–5667, 2018

    2018

  3. [11]

    Zero shot learning via low-rank embedded semantic autoencoder.,

    Y . Liu, Q. Gao, J. Li, J. Han, and L. Shao, “Zero shot learning via low-rank embedded semantic autoencoder.,” in International Joint Conference on Artificial Intelligence , pp. 2490–2496, 2018

    2018

  4. [12]

    Generalized zero-shot recognition based on visually semantic embedding,

    P. Zhu, H. Wang, and V . Saligrama, “Generalized zero-shot recognition based on visually semantic embedding,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, in press, 2019

    2019

  5. [13]

    Generalized zero- shot learning with deep calibration network,

    S. Liu, M. Long, J. Wang, and M. I. Jordan, “Generalized zero- shot learning with deep calibration network,” in Neural Information Processing Systems, 2018

    2018

  6. [14]

    Gener- alized zero-and few-shot learning via aligned variational autoencoders,

    E. Sch ¨onfeld, S. Ebrahimi, S. Sinha, T. Darrell, and Z. Akata, “Gener- alized zero-and few-shot learning via aligned variational autoencoders,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, in press , 2019

    2019

  7. [15]

    Feature generating networks for zero-shot learning,

    Y . Xian, T. Lorenz, B. Schiele, and Z. Akata, “Feature generating networks for zero-shot learning,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2018

    2018

  8. [16]

    Multi-modal cycle- consistent generalized zero-shot learning,

    R. Felix, V . B. Kumar, I. Reid, and G. Carneiro, “Multi-modal cycle- consistent generalized zero-shot learning,” in European Conference on Computer Vision, pp. 21–37, 2018

    2018

  9. [17]

    Zero-shot kernel learning,

    H. Zhang and P. Koniusz, “Zero-shot kernel learning,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , June 2018

    2018

  10. [18]

    Nips 2016 tutorial: Generative adversarial networks,

    I. Goodfellow, “Nips 2016 tutorial: Generative adversarial networks,” arXiv:1701.00160, 2016

    2016 arXiv

  11. [19]

    Learning to detect unseen object classes by between-class attribute transfer,

    C. H. Lampert, H. Nickisch, and S. Harmeling, “Learning to detect unseen object classes by between-class attribute transfer,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2009

    2009

  12. [20]

    Attribute-based classifi- cation for zero-shot visual object categorization,

    C. H. Lampert, H. Nickisch, and S. Harmeling, “Attribute-based classifi- cation for zero-shot visual object categorization,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 36, no. 3, pp. 453–465, 2013

    2013

  13. [21]

    Semi-supervised zero-shot classi- fication with label representation learning,

    X. Li, Y . Guo, and D. Schuurmans, “Semi-supervised zero-shot classi- fication with label representation learning,” in Proceedings of the IEEE International Conference on Computer Vision , pp. 4211–4219, 2015

    2015

  14. [22]

    Zero-shot learning-the good, the bad and the ugly,

    Y . Xian, B. Schiele, and Z. Akata, “Zero-shot learning-the good, the bad and the ugly,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pp. 4582–4591, 2017

    2017

  15. [23]

    Recent advances in zero-shot recognition: Toward data-efficient understanding of visual content,

    Y . Fu, T. Xiang, Y .-G. Jiang, X. Xue, L. Sigal, and S. Gong, “Recent advances in zero-shot recognition: Toward data-efficient understanding of visual content,” IEEE Signal Processing Magazine , vol. 35, no. 1, pp. 112–125, 2018

    2018

  16. [24]

    Zero-shot learning: An energy based approach,

    T. Zhao, G. Liu, L. Wu, C. Ma, and E. Chen, “Zero-shot learning: An energy based approach,” in IEEE International Conference on Data Mining, 2018

    2018

  17. [25]

    Learning to share visual appearance for multiclass object detection,

    R. Salakhutdinov, A. Torralba, and J. Tenenbaum, “Learning to share visual appearance for multiclass object detection,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition

  18. [26]

    Capturing long-tail distribu- tions of object subcategories,

    X. Zhu, D. Anguelov, and D. Ramanan, “Capturing long-tail distribu- tions of object subcategories,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pp. 915–922, 2014

    2014

  19. [27]

    Zero-shot learning via semantic similarity embedding,

    Z. Zhang and V . Saligrama, “Zero-shot learning via semantic similarity embedding,” in Proceedings of the IEEE International Conference on Computer Vision, pp. 4166–4174, 2015

    2015

  20. [28]

    An embarrassingly simple approach to zero-shot learning,

    B. Romera-Paredes and P. Torr, “An embarrassingly simple approach to zero-shot learning,” in International Conference on Machine Learning , pp. 2152–2161, 2015

    2015

  21. [29]

    Latent embeddings for zero-shot classification,

    Y . Xian, Z. Akata, G. Sharma, Q. Nguyen, M. Hein, and B. Schiele, “Latent embeddings for zero-shot classification,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pp. 69– 77, 2016

    2016

  22. [30]

    Semantic autoencoder for zero-shot learning,

    E. Kodirov, T. Xiang, and S. Gong, “Semantic autoencoder for zero-shot learning,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 3174–3183, 2017

    2017

  23. [31]

    Learning a deep embedding model for zero-shot learning,

    L. Zhang, T. Xiang, and S. Gong, “Learning a deep embedding model for zero-shot learning,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pp. 2021–2030, 2017

    2021

  24. [32]

    Implicit non-linear similarity scoring for recognizing unseen classes,

    Y . Guo, G. Ding, J. Han, S. Zhao, and B. Wang, “Implicit non-linear similarity scoring for recognizing unseen classes,” in International Joint Conference on Artificial Intelligence , pp. 4898–4904, 2018

    2018

  25. [33]

    Zero-shot visual recognition using semantics-preserving adversarial embedding network,

    L. Chen, H. Zhang, J. Xiao, W. Liu, and S.-F. Chang, “Zero-shot visual recognition using semantics-preserving adversarial embedding network,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 1043–1052, 2018

    2018

  26. [34]

    Towards effective deep embedding for zero-shot learning,

    L. Zhang, P. Wang, L. Liu, C. Shen, W. Wei, Y . Zhang, and A. V . D. Hengel, “Towards effective deep embedding for zero-shot learning,” arXiv:1808.10075, 2018

    2018 arXiv

  27. [35]

    Preserving semantic relations for zero-shot learning,

    Y . Annadani and S. Biswas, “Preserving semantic relations for zero-shot learning,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 7603–7612, 2018

    2018

  28. [36]

    Dual-verification network for zero-shot learning,

    H. Zhang, Y . Long, W. Yang, and L. Shao, “Dual-verification network for zero-shot learning,” Information Sciences, vol. 470, pp. 43–57, 2019

    2019

  29. [37]

    Dissimilarity representation learning for generalized zero-shot recognition,

    G. Yang, J. Liu, J. Xu, and X. Li, “Dissimilarity representation learning for generalized zero-shot recognition,” in ACM Conference on Multime- dia, pp. 2032–2039, 2018

    2018

  30. [38]

    Pseudo transfer with marginalized corrupted attribute for zero-shot learning,

    T. Long, X. Xu, Y . Li, F. Shen, J. Song, and H. T. Shen, “Pseudo transfer with marginalized corrupted attribute for zero-shot learning,” in ACM Multimedia Conference on Multimedia Conference , pp. 1802– 1810, 2018

    2018

  31. [39]

    Generalized zero- shot learning via synthesized examples,

    V . K. Verma, G. Arora, A. Mishra, and P. Rai, “Generalized zero- shot learning via synthesized examples,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pp. 4281– 4289, 2018

    2018

  32. [40]

    Bi-adversarial auto-encoder for zero-shot learning,

    Y . Yu, Z. Ji, Y . Pang, J. Guo, Z. Zhang, and F. Wu, “Bi-adversarial auto-encoder for zero-shot learning,” arXiv:1811.08103, 2018

    2018 arXiv

  33. [41]

    Leveraging the invariant side of generative zero-shot learning,

    J. Li, M. Jin, K. Lu, Z. Ding, L. Zhu, and Z. Huang, “Leveraging the invariant side of generative zero-shot learning,”Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, in press, 2019

    2019

  34. [42]

    Multi-modal ensemble classification for generalized zero shot learning,

    R. Felix, M. Sasdelli, I. Reid, and G. Carneiro, “Multi-modal ensemble classification for generalized zero shot learning,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, in press, 2019

    2019

  35. [43]

    Learning to separate domains in generalized zero-shot and open set learning: a probabilistic perspective,

    H. Dong, Y . Fu, L. Sigal, S. J. Hwang, Y .-G. Jiang, and X. Xue, “Learning to separate domains in generalized zero-shot and open set learning: a probabilistic perspective,” arXiv:1810.07368, 2018

    2018 arXiv

  36. [44]

    Adaptive confidence smoothing for gen- eralized zero-shot learning,

    Y . Atzmon and G. Chechik, “Adaptive confidence smoothing for gen- eralized zero-shot learning,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, in press , 2019

    2019

  37. [45]

    Zero-shot learning- a comprehensive evaluation of the good, the bad and the ugly,

    Y . Xian, C. H. Lampert, B. Schiele, and Z. Akata, “Zero-shot learning- a comprehensive evaluation of the good, the bad and the ugly,” IEEE Transactions on Pattern Analysis and Machine Intelligence , 2018

    2018

  38. [46]

    The caltech-ucsd birds-200-2011 dataset,

    C. Wah, S. Branson, P. Welinder, P. Perona, and S. Belongie, “The caltech-ucsd birds-200-2011 dataset,” 2011

    2011

  39. [47]

    Sun attribute database: Discovering, annotat- ing, and recognizing scene attributes,

    G. Patterson and J. Hays, “Sun attribute database: Discovering, annotat- ing, and recognizing scene attributes,” in IEEE Conference on Computer Vision and Pattern Recognition , pp. 2751–2758, 2012

    2012

  40. [48]

    Describing objects by their attributes,

    A. Farhadi, I. Endres, D. Hoiem, and D. Forsyth, “Describing objects by their attributes,” in IEEE Conference on Computer Vision and Pattern Recognition, pp. 1778–1785, 2009

    2009

  41. [49]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 770–778, 2016

    2016

  42. [50]

    Imagenet large scale visual recognition challenge,

    O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh, S. Ma, Z. Huang, A. Karpathy, A. Khosla, M. Bernstein, et al., “Imagenet large scale visual recognition challenge,” International journal of computer vision, vol. 115, no. 3, pp. 211–252, 2015

    2015

  43. [51]

    Learning to compare: Relation network for few-shot learning,

    F. Sung, Y . Yang, L. Zhang, T. Xiang, P. H. Torr, and T. M. Hospedales, “Learning to compare: Relation network for few-shot learning,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 1199–1208, 2018

    2018

  44. [52]

    Open-world visual recognition using knowledge graphs,

    V . Lonij, A. Rawat, and M.-I. Nicolae, “Open-world visual recognition using knowledge graphs,” arXiv preprint arXiv:1708.08310 , 2017

    2017 arXiv

  45. [53]

    Inference with the universum,

    J. Weston, R. Collobert, F. Sinz, L. Bottou, and V . Vapnik, “Inference with the universum,” inProceedings of the 23rd international conference on Machine learning , pp. 1009–1016, ACM, 2006

    2006

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.