Pith. sign in

REVIEW 5 major objections 6 minor 63 references

Compositional Attribute Imbalance in Vision Datasets

T0 review · 5 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Image classifiers fail systematically on rare attribute combinations, and a scarcity-weighted sampler fixes most of the gap.

desk verdict The CAS attribute-scarcity measurement is a genuinely useful idea, but Table 2's internal arithmetic is broken, so the reported gains are unsupported as printed. read the letter →

arxiv 2506.14418 v1 pith:4SWVCVSM submitted 2025-06-17 cs.CV cs.AI

Jiayi Chen , Yanbiao Ma , Andi Zhang , Weidong Tang , Wei Dai , Bowei Liu This is my paper · ORCID
classification cs.CVcs.AI
keywords compositionalattributeimbalancevisualdictionaryCLIPscarcitydataaugmentationlong-tailclassificationsamplingstrategyimage
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

Vision models are usually trained to balance classes, but this paper argues that a second, hidden imbalance matters more: within a single class, rare combinations of visual attributes such as a white dog or a striped car are learned much worse than common ones. To measure this, the authors use CLIP to build a visual attribute dictionary and assign each image 20 attribute labels, then define Compositional Attribute Scarcity (CAS) as the sum of the within-class frequency ranks of those attributes. Across 12 datasets, accuracy falls steadily as CAS rises, with the worst accuracy concentrated on the rarest attribute combinations. The proposed remedy is a pure sampling change: draw each image with probability proportional to its CAS raised to the power 1.2, and feed those batches into standard augmentation methods. The paper reports accuracy gains on every dataset, with the largest gains exactly on the high-CAS samples that were previously worst.

What carries the argument

The load-bearing object is the Compositional Attribute Scarcity score (CAS), defined as the sum, over an image's 20 CLIP-assigned secondary attributes, of the within-class frequency ranks of those attributes (rank 1 means most frequent). Constructing it requires the paper's visual attribute dictionary: CLIP text embeddings of phrases such as "the photo is Brown" are matched to image embeddings to build a key-value dictionary, and each image's embedding is later matched to the nearest key to retrieve its attributes. CAS converts that dictionary into one scalar per sample, and the proposed fix then uses $p_i = r_i^b / \sum_k r_k^b$ with $b=1.2$ as the sampling probability, feeding rare-attribute images more often into mixup-style data augmentation.

What would settle it

Take one dataset such as CIFAR-100 or a fine-grained car dataset, have human annotators label the same 20 attributes on a sample of test images, recompute CAS from those labels, and re-plot accuracy by CAS. If the accuracy decline disappears or reverses, the central claim fails; if it survives, the CLIP-based measurement is not the source of the effect.

Watch

Extended reading notes

Core claim

The paper's central claim is that image classifiers suffer from a distinct, measurable form of data imbalance that is not captured by class counts: compositional attribute imbalance. Using a CLIP-based visual attribute dictionary, each image is annotated with 20 secondary attributes, one per primary attribute category such as color, material, or shape, and the within-class frequency rank of each attribute gives its scarcity. The Compositional Attribute Scarcity (CAS) of an image is the sum of those 20 ranks, so a higher CAS marks an image whose attribute combination is rare inside its own class. Across 12 benchmark datasets and two architectures, the paper reports that accuracy drops monotonically as CAS increases, with the steepest losses on the most scarce samples. It then shows that upsampling high-CAS images, using sampling probability proportional to $r_i^b$ with $b=1.2$ before feeding batches to CutMix, FMix, or SaliencyMix, improves accuracy on every dataset tested, with the largest gains on exactly the high-CAS samples that were previously worst.

Load-bearing premise

The analysis assumes the CLIP attribute dictionary labels images accurately on all twelve datasets, including fine-grained ones, yet the paper never checks those automatic labels against human judgment.

Editorial extensions

If this is right

  • Training pipelines that only balance classes still leave an accuracy gap on images with rare attribute combinations, so class-level balance is not sufficient for fair performance.
  • Swapping in the CAS-weighted sampler, with no extra parameters and no inference cost, improves overall accuracy on all 12 datasets tested and improves high-CAS accuracy the most.
  • The largest gains appear in fine-grained domains such as pets, cars, and flowers, where attribute combinations are the main discriminative signal.
  • The sampler composes with CutMix, FMix, and SaliencyMix, indicating that the benefit is not tied to one augmentation recipe.
  • Because CAS scores are computed offline, the method can be added to existing long-tail training frameworks without retraining an attribute model.

Reading between the lines

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

  • Beyond the paper: CAS could be computed from any pretrained attribute vocabulary, so the same weighting idea could transfer to region-level tasks such as detection or segmentation if per-region attributes were available.
  • Beyond the paper: replacing the sampler with a CAS-weighted loss would separate the effect of seeing rare images more often from the effect of larger gradients on them.
  • Beyond the paper: long-tailed methods that rebalance classes could be composed with CAS rebalancing to handle two independent axes of skew.
  • Beyond the paper: the accuracy-CAS curve is measured with the same CLIP annotations that define CAS, so a control with randomized CAS assignments is needed to rule out annotation noise as the source of the trend.
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

5 major / 6 minor

Summary. This paper studies intra-class attribute imbalance in image classification. The authors define 20 primary attribute categories and 300+ secondary attributes, build a CLIP-based visual attribute dictionary from ImageNet-21k, and use it to annotate 12 benchmark datasets. They propose Compositional Attribute Scarcity (CAS), defined as the sum of within-class frequency ranks of an image's 20 secondary attributes, and report that higher CAS is associated with lower classification accuracy. To mitigate this, they reweight the training sampler by a power of each sample's CAS (r_i^b, with b=1.2) and combine this weighted sampler with CutMix, FMix, and SaliencyMix. Section 5 presents per-subset accuracies (low/middle/high CAS) and overall accuracy, claiming universal improvements across all 12 datasets and three augmentation methods.

Significance. If the reported results were correct, the paper would make a useful contribution: a scalable, model-agnostic way to characterize intra-class attribute imbalance and a zero-overhead sampling modification that consistently improves strong augmentations on a broad range of tasks. The framework is clearly motivated, the experimental scope is wide, and the idea of guiding data augmentation by compositional rarity is interesting. However, the paper's central quantitative evidence is internally inconsistent, the CLIP-based dictionary is not validated, and the method is not compared with generic reweighting baselines. The paper also provides no code, data, or error bars. As a result, the claimed empirical findings are not established in this version.

major comments (5)
  1. [Table 2, §5.3] The 'all' accuracy must equal 0.4×high + 0.3×middle + 0.3×low under the stated 40/30/30 partition, but this identity fails for many rows. For example, DTD with ResNeXt-50 and CutMix reports low=95.22, middle=94.36, high=86.32, yet all=99.61, which is larger than every subset accuracy and therefore arithmetically impossible. CIFAR-100 with ResNet-18 and CutMix reports low=68.31, middle=49.65, high=43.05, giving a weighted average of 52.61, while the table says all=54.57. Because Section 5.5's conclusion that performance improved on all datasets is read directly from these numbers, the central empirical claim of the paper is not supported by the tabulated data.
  2. [§3.4, §5.3] The paper does not state whether the within-class frequency ranks used to define CAS are computed on the training split, the test split, or the entire dataset. If the test set contributes to the rank statistics, then both the test-set subset definition and the training sampler are informed by test-set statistics, which is a leakage concern. The authors must specify the split used and, ideally, recompute all results with ranks estimated from training data only.
  3. [§3.2] The CLIP-based attribute dictionary is constructed once on ImageNet-21k and then applied unmodified to all 12 datasets, including fine-grained domains such as Stanford Cars, CUB-200-2011, Oxford-102 Flower, and DTD. No validation against human attribute labels, no per-dataset transfer check, and no measure of annotation noise are provided. Since every CAS score and every subset in Table 2 depends on this dictionary, the whole analysis is unreliable until the dictionary is shown to be accurate on the target domains.
  4. [§4.1, §5.5] The training sampler upweights samples by the same CAS scores that define the low/middle/high test subsets, so an increase in high-CAS accuracy is partly a mechanical consequence of training more often on high-CAS samples. The paper does not compare with generic reweighting baselines such as class-balanced sampling, inverse-frequency sampling, or loss-based hard-example mining. Without such comparisons, the claimed CAS-specific benefit is not established.
  5. [§5.4, §5.5] The power parameter b=1.2 is selected on CIFAR-100 and ImageNet, which are also part of the main evaluation, and all results are point estimates with no standard deviations or repeated runs. Given that Table 2 already contains arithmetically impossible entries, the reported gains—up to 10.79 points on Stanford Cars—cannot be evaluated for statistical significance.
minor comments (6)
  1. [Algorithm 1, §4.1] Algorithm 1 uses r_i^β in the weight update, while Section 4.1 defines the transformation as r_i^b; the notation should be unified. Additionally, the algorithm input line says 'Combination rarity scores' rather than 'compositional attribute scarcity scores.'
  2. [§3.3, Figure 2] The text says the analysis uses ResNet-18 and ResNet-50, while the Figure 2 caption mentions ResNeXt-50; the architecture names should be made consistent throughout.
  3. [§5.1] Stanford Dogs is described as containing 20 dog breeds, but the actual dataset contains 120 breeds; this description should be corrected.
  4. [Figure 4] The description of the horizontal axis as '10 evenly divided intervals' does not explain how CAS values are mapped to intervals, and the right vertical axis is not labeled; the figure should be self-contained.
  5. [Reproducibility] The paper does not release the attribute dictionary, per-image CAS scores, or code, so the numbers in Figures 2–4 and Table 2 cannot be reproduced or independently checked.
  6. [Abstract and §3.1] The abstract uses 'first-level and second-level attributes' while the body uses 'primary and secondary attributes'; terminology should be made consistent.

Circularity Check

1 steps flagged · score 2.0 of 10

No definitional circularity in the CAS–accuracy analysis; the only circular-adjacent step is that the power parameter b is tuned on CIFAR-100/ImageNet and those same datasets are then highlighted in the main results.

  1. fitted input called prediction [Section 5.4 (Selection of Hyperparameter b), Section 5.5, and Table 2 CIFAR-100/ImageNet-1k rows]
    "We explored the optimal value of b by setting it within the range of 0.5 to 1.5 on CIFAR-100 and ImageNet. As shown in Figure 5, when b=1.2, our method achieves the highest performance gains for CutMix, FMix, and SaliencyMix. Therefore, we set b=1.2 for all subsequent experiments."

    The only free parameter of the proposed sampler, b, is selected to maximize gains on CIFAR-100 and ImageNet, and those same two datasets appear in Table 2 and are called out in Section 5.5 (e.g., 'on ImageNet-1k ... improved ... by 1.18%, 1.58%, and 3.07%'). Thus the reported improvements on these rows are in-sample with respect to hyperparameter selection rather than independent predictions. This is a selection/leakage caveat, not a definitional reduction: the CAS-accuracy correlation is defined independently of b, and ten other datasets are not used in tuning.

full rationale

The paper's central quantity, CAS, is the sum of within-class frequency ranks of secondary attributes assigned by a CLIP-based dictionary; it is not defined in terms of model accuracy or of the reweighting probabilities. The observed monotone decrease of accuracy with CAS is therefore an empirical correlation, not a tautology. The reweighting scheme p_i proportional to r_i^b changes the training distribution, and the resulting accuracy changes on CAS-stratified test subsets are a measurable outcome rather than an identity. Self-citations (Ma et al.) appear only in related-work framing and are not load-bearing. A separate serious concern is that several Table 2 rows have an 'all' column that cannot equal 0.4*High + 0.3*Middle + 0.3*Low under the Section 5.3 partition, e.g., DTD ResNeXt-50 CutMix reports all=99.61 while its three subsets are 95.22/94.36/86.32; this is an internal-consistency and reproducibility problem, not a circularity. Overall, the derivation is not circular; the only circular-adjacent issue is the in-sample b selection on two datasets, which prevents a clean score of 0.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

All load-bearing components of the paper are empirical modeling choices that are not validated against external benchmarks. The only fitted numeric parameter is the power b, tuned on CIFAR-100 and ImageNet and then reused on those same datasets in the main results. The CLIP dictionary, the attribute taxonomy, and the scarcity definition are unverified domain assumptions.

free parameters (2)
  • b (CAS power parameter) = 1.2
    Tuned on CIFAR-100 and ImageNet in Section 5.4; controls the nonlinear amplification of CAS in the sampling weight r_i^b.
  • Number of primary attributes = 20
    Hand-selected taxonomy from prior work (Zhong et al., 2021; Zhang et al., 2024); no criterion given for completeness or coverage across the 12 datasets.
assumptions (5)
  • domain assumption CLIP text-image similarity is a valid proxy for visual attribute presence.
    The attribute dictionary and all CAS annotations rely on CLIP cosine similarity; no human or ground-truth validation is provided (Section 3.2).
  • domain assumption The 20 primary and 300+ secondary attribute taxonomy is comprehensive across the 12 datasets.
    Taxonomy is inherited from prior studies but not justified for fine-grained and texture datasets (Section 3.1).
  • domain assumption Within-class frequency rank of a secondary attribute is a meaningful measure of scarcity.
    CAS sums these ranks; the paper does not justify why rank position, rather than frequency or entropy, captures difficulty (Section 3.4).
  • domain assumption The ImageNet-21k attribute dictionary transfers to all target datasets without domain shift.
    Dictionary keys are image prototypes from ImageNet-21k only; no adaptation or transfer validation is reported (Section 3.2).
  • domain assumption Reweighting the sampler by r_i^b improves rare-attribute learning without hurting common attributes.
    This is the design hypothesis of the method; the paper tests it empirically but the mechanism is not derived (Section 4.1).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Compositional Attribute Imbalance in Vision Datasets." pith.science (2026). https://pith.science/paper/4SWVCVSM

@misc{pith2026250614418,
  author       = {Pith},
  title        = {Pith review of: Compositional Attribute Imbalance in Vision Datasets},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4SWVCVSM}},
  note         = {Machine review of arXiv:2506.14418}
}
read the original abstract

Visual attribute imbalance is a common yet underexplored issue in image classification, significantly impacting model performance and generalization. In this work, we first define the first-level and second-level attributes of images and then introduce a CLIP-based framework to construct a visual attribute dictionary, enabling automatic evaluation of image attributes. By systematically analyzing both single-attribute imbalance and compositional attribute imbalance, we reveal how the rarity of attributes affects model performance. To tackle these challenges, we propose adjusting the sampling probability of samples based on the rarity of their compositional attributes. This strategy is further integrated with various data augmentation techniques (such as CutMix, Fmix, and SaliencyMix) to enhance the model's ability to represent rare attributes. Extensive experiments on benchmark datasets demonstrate that our method effectively mitigates attribute imbalance, thereby improving the robustness and fairness of deep neural networks. Our research highlights the importance of modeling visual attribute distributions and provides a scalable solution for long-tail image classification tasks.

Figures

Figures reproduced from arXiv: 2506.14418 by the authors.

Figure 1
Figure 1. The left side shows all primary attributes we defined and their corresponding secondary attributes. The right side illustrates the process of constructing the visual attribute dictionary based on CLIP. 3. Attribute Imbalance In this section, we first systematically define the visual attributes of images. Then, we propose using CLIP to con￾struct a visual attribute dictionary, enabling automatic eval￾uation of image … view at source ↗
Figure 2
Figure 2. The distribution of secondary attributes under color and material categories across 12 visual benchmark datasets, along with the performance of ResNet-18 and ResNeXt-50 on each secondary attribute. model’s performance on low-frequency attribute samples. Considering that an image often contains multiple visual attributes, we further introduce the concept of compositional attributes to explore the impact of multi-attr… view at source ↗
Figure 3
Figure 3. illustrates the process of calculating the CAS of an … … Color Frequencies 1 2 3 4 5 red … Texture Frequencies 1 2 3 4 5 Striped CAS=3+…+2 [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: The long-tailed distribution of sample composite attribute sparsity across certain categories in 12 visual benchmark datasets, along with the performance of ResNet-18 and ResNeXt-50 across different compositional attribute scarcity (CAS) intervals. The horizontal axis …
Figure 5
Figure 5. Figure 5: Performance of ResNeXt-50 with our method combined with CutMix, Fmix, and SaliencyMix under different values of b. sparsity-based subset division allows us to more precisely analyze the model’s performance under varying informa￾tion conditions, particularly in terms of…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

63 extracted references · 39 canonical work pages

  1. [1]

    Food-101--mining discriminative components with random forests

    Bossard, L., Guillaumin, M., and Van Gool, L. Food-101--mining discriminative components with random forests. In Computer vision--ECCV 2014: 13th European conference, zurich, Switzerland, September 6-12, 2014, proceedings, part VI 13, pp.\ 446--461. Springer, 2014

    2014

  2. [2]

    Ace: Ally complementary experts for solving long-tailed recognition in one-shot

    Cai, J., Wang, Y., and Hwang, J.-N. Ace: Ally complementary experts for solving long-tailed recognition in one-shot. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp.\ 112--121, 2021

    work page 2021

  3. [3]

    Learning imbalanced datasets with label-distribution-aware margin loss

    Cao, K., Wei, C., Gaidon, A., Arechiga, N., and Ma, T. Learning imbalanced datasets with label-distribution-aware margin loss. Advances in Neural Information Processing Systems, 32, 2019

    work page 2019

  4. [4]

    V., Bowyer, K

    Chawla, N. V., Bowyer, K. W., Hall, L. O., and Kegelmeyer, W. P. Smote: synthetic minority over-sampling technique. Journal of Artificial Intelligence Research, 16: 0 321--357, 2002

    2002

  5. [5]

    Feature space augmentation for long-tailed data

    Chu, P., Bian, X., Liu, S., and Ling, H. Feature space augmentation for long-tailed data. In European Conference on Computer Vision, pp.\ 694--710. Springer, 2020

    work page 2020

  6. [6]

    Describing textures in the wild

    Cimpoi, M., Maji, S., Kokkinos, I., Mohamed, S., and Vedaldi, A. Describing textures in the wild. In Proceedings of the IEEE conference on computer vision and pattern recognition, pp.\ 3606--3613, 2014

    2014

  7. [7]

    Parametric contrastive learning

    Cui, J., Zhong, Z., Liu, S., Yu, B., and Jia, J. Parametric contrastive learning. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp.\ 715--724, 2021

    2021

  8. [8]

    Large scale fine-grained categorization and domain-specific transfer learning

    Cui, Y., Song, Y., Sun, C., Howard, A., and Belongie, S. Large scale fine-grained categorization and domain-specific transfer learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 4109--4118, 2018

    work page 2018

Show all 63 references
  1. [9]

    Class-balanced loss based on effective number of samples

    Cui, Y., Jia, M., Lin, T.-Y., Song, Y., and Belongie, S. Class-balanced loss based on effective number of samples. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 9268--9277, 2019

    2019

  2. [10]

    Imagenet: A large-scale hierarchical image database

    Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., and Fei-Fei, L. Imagenet: A large-scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition, pp.\ 248--255. Ieee, 2009

    2009

  3. [11]

    Class rectification hard mining for imbalanced deep learning

    Dong, Q., Gong, S., and Zhu, X. Class rectification hard mining for imbalanced deep learning. In Proceedings of the IEEE International Conference on Computer Vision, pp.\ 1851--1860, 2017

    2017

  4. [12]

    H., Williams, K., Corke, F

    Doonan, J. H., Williams, K., Corke, F. M., Zhang, H., Liu, Y., et al. Handling intra-class imbalance in part-segmentation of different wheat cultivars. Computers and Electronics in Agriculture, 230: 0 109826, 2025

    2025

  5. [13]

    The foundations of cost-sensitive learning

    Elkan, C. The foundations of cost-sensitive learning. In International joint conference on artificial intelligence, volume 17, pp.\ 973--978. Lawrence Erlbaum Associates Ltd, 2001

    2001

  6. [14]

    A multiple resampling method for learning from imbalanced data sets

    Estabrooks, A., Jo, T., and Japkowicz, N. A multiple resampling method for learning from imbalanced data sets. Computational intelligence, 20 0 (1): 0 18--36, 2004

    2004

  7. [15]

    Learning to segment the tail

    Hu, X., Jiang, Y., Tang, K., Chen, J., Miao, C., and Zhang, H. Learning to segment the tail. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 14045--14054, 2020

    2020

  8. [16]

    C., and Tang, X

    Huang, C., Li, Y., Loy, C. C., and Tang, X. Learning deep representation for imbalanced classification. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 5375--5384, 2016

    2016

  9. [17]

    Z., Mahmood, A., and Nandakumar, K

    Islam, K., Zaheer, M. Z., Mahmood, A., and Nandakumar, K. Diffusemix: Label-preserving data augmentation with diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 27621--27630, 2024

    2024

  10. [18]

    Exploring balanced feature spaces for representation learning

    Kang, B., Li, Y., Xie, S., Yuan, Z., and Feng, J. Exploring balanced feature spaces for representation learning. In International Conference on Learning Representations, 2020

    2020

  11. [19]

    Novel dataset for fine-grained image categorization: Stanford dogs

    Khosla, A., Jayadevaprakash, N., Yao, B., and Li, F.-F. Novel dataset for fine-grained image categorization: Stanford dogs. In Proc. CVPR workshop on fine-grained visual categorization (FGVC), volume 2, 2011

    2011

  12. [20]

    3d object representations for fine-grained categorization

    Krause, J., Stark, M., Deng, J., and Fei-Fei, L. 3d object representations for fine-grained categorization. In Proceedings of the IEEE international conference on computer vision workshops, pp.\ 554--561, 2013

    2013

  13. [21]

    Learning multiple layers of features from tiny images

    Krizhevsky, A., Hinton, G., et al. Learning multiple layers of features from tiny images. 2009

    2009

  14. [22]

    Focal loss for dense object detection

    Lin, T.-Y., Goyal, P., Girshick, R., He, K., and Doll \'a r, P. Focal loss for dense object detection. In Proceedings of the IEEE International Conference on Computer Vision, pp.\ 2980--2988, 2017

    2017

  15. [23]

    Gistnet: a geometric structure transfer network for long-tailed recognition

    Liu, B., Li, H., Kang, H., Hua, G., and Vasconcelos, N. Gistnet: a geometric structure transfer network for long-tailed recognition. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp.\ 8209--8218, 2021 a

    2021

  16. [24]

    Deep representation learning on long-tailed data: A learnable embedding augmentation perspective

    Liu, J., Sun, Y., Han, C., Dou, Z., and Li, W. Deep representation learning on long-tailed data: A learnable embedding augmentation perspective. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 2970--2979, 2020

    2020

  17. [25]

    Deepfashion: Powering robust clothes recognition and retrieval with rich annotations

    Liu, Z., Luo, P., Qiu, S., Wang, X., and Tang, X. Deepfashion: Powering robust clothes recognition and retrieval with rich annotations. In Proceedings of the IEEE conference on computer vision and pattern recognition, pp.\ 1096--1104, 2016

    2016

  18. [26]

    Handling inter-class and intra-class imbalance in class-imbalanced learning

    Liu, Z., Wei, P., Wei, Z., Yu, B., Jiang, J., Cao, W., Bian, J., and Chang, Y. Handling inter-class and intra-class imbalance in class-imbalanced learning. arXiv preprint arXiv:2111.12791, 2021 b

    2021 arXiv

  19. [27]

    Delving into semantic scale imbalance

    Ma, Y., Jiao, L., Liu, F., Li, Y., Yang, S., and Liu, X. Delving into semantic scale imbalance. In The Eleventh International Conference on Learning Representations, 2023. URL https://openreview.net/forum?id=07tc5kKRIo

    2023

  20. [28]

    Geometric prior guided feature representation learning for long-tailed classification

    Ma, Y., Jiao, L., Liu, F., Yang, S., Liu, X., and Chen, P. Geometric prior guided feature representation learning for long-tailed classification. International Journal of Computer Vision, pp.\ 1--18, 2024 a

    2024

  21. [29]

    Feature distribution representation learning based on knowledge transfer for long-tailed classification

    Ma, Y., Jiao, L., Liu, F., Yang, S., Liu, X., and Chen, P. Feature distribution representation learning based on knowledge transfer for long-tailed classification. IEEE Transactions on Multimedia, 26: 0 2772--2784, 2024 b . doi:10.1109/TMM.2023.3303697

    2024

  22. [30]

    Predicting and enhancing the fairness of dnns with the curvature of perceptual manifolds

    Ma, Y., Jiao, L., Liu, F., Wen, M., Li, L., Ma, W., Yang, S., Liu, X., and Chen, P. Predicting and enhancing the fairness of dnns with the curvature of perceptual manifolds. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025

    2025

  23. [31]

    Fine-grained visual classification of aircraft

    Maji, S., Rahtu, E., Kannala, J., Blaschko, M., and Vedaldi, A. Fine-grained visual classification of aircraft. arXiv preprint arXiv:1306.5151, 2013

    2013 arXiv

  24. [32]

    and Zisserman, A

    Nilsback, M.-E. and Zisserman, A. Automated flower classification over a large number of classes. In 2008 Sixth Indian conference on computer vision, graphics & image processing, pp.\ 722--729. IEEE, 2008

    2008

  25. [33]

    Factors in finetuning deep model for object detection with long-tail distribution

    Ouyang, W., Wang, X., Zhang, C., and Yang, X. Factors in finetuning deep model for object detection with long-tail distribution. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 864--873, 2016

    2016

  26. [34]

    Park, S., Hong, Y., Heo, B., Yun, S., and Choi, J. Y. The majority can help the minority: Context-rich minority oversampling for long-tailed classification. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 6887--6896, 2022

    2022

  27. [35]

    M., Vedaldi, A., Zisserman, A., and Jawahar, C

    Parkhi, O. M., Vedaldi, A., Zisserman, A., and Jawahar, C. Cats and dogs. In 2012 IEEE conference on computer vision and pattern recognition, pp.\ 3498--3505. IEEE, 2012

    2012

  28. [36]

    Learning to predict visual attributes in the wild

    Pham, K., Kafle, K., Lin, Z., Ding, Z., Cohen, S., Tran, Q., and Shrivastava, A. Learning to predict visual attributes in the wild. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp.\ 13018--13028, 2021

    2021

  29. [37]

    A., and Gao, X

    Qin, H., Jin, X., Zhu, H., Liao, H., El-Yacoubi, M. A., and Gao, X. Sumix: Mixup with semantic and uncertain information. In European Conference on Computer Vision, pp.\ 70--88. Springer, 2024

    2024

  30. [38]

    W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., et al

    Radford, A., Kim, J. W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., et al. Learning transferable visual models from natural language supervision. In International conference on machine learning, pp.\ 8748--8763. PMLR, 2021

    2021

  31. [39]

    Balanced meta-softmax for long-tailed visual recognition

    Ren, J., Yu, C., Ma, X., Zhao, H., Yi, S., et al. Balanced meta-softmax for long-tailed visual recognition. Advances in Neural Information Processing Systems, 33: 0 4175--4186, 2020

    2020

  32. [40]

    Class-wise difficulty-balanced loss for solving class-imbalance

    Sinha, S., Ohashi, H., and Nakamura, K. Class-wise difficulty-balanced loss for solving class-imbalance. In Proceedings of the Asian Conference on Computer Vision, 2020

    2020

  33. [41]

    Class-difficulty based methods for long-tailed visual recognition

    Sinha, S., Ohashi, H., and Nakamura, K. Class-difficulty based methods for long-tailed visual recognition. International Journal of Computer Vision, 130 0 (10): 0 2517--2531, 2022

    2022

  34. [42]

    Equalization loss for long-tailed object recognition

    Tan, J., Wang, C., Li, B., Li, Q., Ouyang, W., Yin, C., and Yan, J. Equalization loss for long-tailed object recognition. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 11662--11671, 2020

    2020

  35. [43]

    Invariant feature learning for generalized long-tailed classification

    Tang, K., Tao, M., Qi, J., Liu, Z., and Zhang, H. Invariant feature learning for generalized long-tailed classification. In European Conference on Computer Vision, pp.\ 709--726. Springer, 2022

    2022

  36. [44]

    The caltech-ucsd birds-200-2011 dataset

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

    2011

  37. [45]

    The devil is in classification: A simple framework for long-tail instance segmentation

    Wang, T., Li, Y., Kang, B., Li, J., Liew, J., Tang, S., Hoi, S., and Feng, J. The devil is in classification: A simple framework for long-tail instance segmentation. In European Conference on Computer Vision, pp.\ 728--744. Springer, 2020 a

    2020

  38. [46]

    Wang, X., Lian, L., Miao, Z., Liu, Z., and Yu, S. X. Long-tailed recognition by routing diverse distribution-aware experts. arXiv preprint arXiv:2010.01809, 2020 b

    2010 arXiv

  39. [47]

    A., Oliva, A., and Torralba, A

    Xiao, J., Hays, J., Ehinger, K. A., Oliva, A., and Torralba, A. Sun database: Large-scale scene recognition from abbey to zoo. In 2010 IEEE computer society conference on computer vision and pattern recognition, pp.\ 3485--3492. IEEE, 2010

    2010

  40. [48]

    Defect spectrum: a granular look of large-scale defect datasets with rich semantics

    Yang, S., Chen, Z., Chen, P., Fang, X., Liang, Y., Liu, S., and Chen, Y. Defect spectrum: a granular look of large-scale defect datasets with rich semantics. In European Conference on Computer Vision, pp.\ 187--203. Springer, 2024

    2024

  41. [49]

    and Xu, Z

    Yang, Y. and Xu, Z. Rethinking the value of labels for improving class-imbalanced learning. Advances in Neural Information Processing Systems, 33: 0 19290--19301, 2020

    2020

  42. [50]

    Identifying and compensating for feature deviation in imbalanced deep learning

    Ye, H.-J., Chen, H.-Y., Zhan, D.-C., and Chao, W.-L. Identifying and compensating for feature deviation in imbalanced deep learning. arXiv preprint arXiv:2001.01385, 2020

    2001 arXiv

  43. [51]

    Feature transfer learning for face recognition with under-represented data

    Yin, X., Yu, X., Sohn, K., Liu, X., and Chandraker, M. Feature transfer learning for face recognition with under-represented data. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 5704--5713, 2019

    2019

  44. [52]

    Zang, Y., Huang, C., and Loy, C. C. Fasa: Feature augmentation and sampling adaptation for long-tailed instance segmentation. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp.\ 3457--3466, 2021

    2021

  45. [53]

    Test-agnostic long-tailed recognition by test-time aggregating diverse experts with self-supervision

    Zhang, Y., Hooi, B., Hong, L., and Feng, J. Test-agnostic long-tailed recognition by test-time aggregating diverse experts with self-supervision. arXiv preprint arXiv:2107.09249, 2021 a

    2021 arXiv

  46. [54]

    Deep long-tailed learning: A survey

    Zhang, Y., Kang, B., Hooi, B., Yan, S., and Feng, J. Deep long-tailed learning: A survey. arXiv preprint arXiv:2110.04596, 2021 b

    2021 arXiv

  47. [55]

    Concept-guided prompt learning for generalization in vision-language models

    Zhang, Y., Zhang, C., Yu, K., Tang, Y., and He, Z. Concept-guided prompt learning for generalization in vision-language models. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pp.\ 7377--7386, 2024

    2024

  48. [56]

    and Pfister, T

    Zhang, Z. and Pfister, T. Learning fast sample re-weighting without reward data. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp.\ 725--734, 2021

    2021

  49. [57]

    A large-scale attribute dataset for zero-shot learning

    Zhao, B., Fu, Y., Liang, R., Wu, J., Wang, Y., and Wang, Y. A large-scale attribute dataset for zero-shot learning. In Proceedings of the ieee/cvf conference on computer vision and pattern recognition workshops, pp.\ 0--0, 2019

    2019

  50. [58]

    C., Tan, M., and Huang, J

    Zhao, P., Zhang, Y., Wu, M., Hoi, S. C., Tan, M., and Huang, J. Adaptive cost-sensitive online classification. IEEE Transactions on Knowledge and Data Engineering, 31 0 (2): 0 214--228, 2018

    2018

  51. [59]

    Improving calibration for long-tailed recognition

    Zhong, Z., Cui, J., Liu, S., and Jia, J. Improving calibration for long-tailed recognition. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 16489--16498, 2021

    2021

  52. [60]

    Bbn: Bilateral-branch network with cumulative learning for long-tailed visual recognition

    Zhou, B., Cui, Q., Wei, X.-S., and Chen, Z.-M. Bbn: Bilateral-branch network with cumulative learning for long-tailed visual recognition. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 9719--9728, 2020

    2020

  53. [61]

    Global-local framework for medical image segmentation with intra-class imbalance problem

    Zhou, Y., Yang, B., Lin, X., Higashita, R., and Liu, J. Global-local framework for medical image segmentation with intra-class imbalance problem. In Proceedings of the 2023 2nd Asia Conference on Algorithms, Computing and Machine Learning, pp.\ 366--370, 2023

    2023

  54. [62]

    and Liu, X.-Y

    Zhou, Z.-H. and Liu, X.-Y. Training cost-sensitive neural networks with methods addressing the class imbalance problem. IEEE Transactions on Knowledge and Data Engineering, 18 0 (1): 0 63--77, 2005

    2005

  55. [63]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.