Pith. sign in

REVIEW 3 major objections 5 minor 54 references

HydraMix: Multi-Image Feature Mixing for Small Data Image Classification

T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read HydraMix generates new training images by mixing an arbitrary number of same-class images in feature space under segmentation-guided masks, and the paper reports that this sets a new state of the art in small-data image classification.

desk verdict HydraMix is a credible incremental extension of ChimeraMix to N-image mixing with a new CLIP-based diversity metric, but the state-of-the-art claim outruns the statistics: most margins over ChimeraMix sit within one standard deviation and hyperparameter selection may have used test accuracy. read the letter →

arxiv 2501.09504 v1 pith:XXSDQE3D submitted 2025-01-16 cs.CV

Christoph Reinders , Frederik Schubert , Bodo Rosenhahn This is my paper · ORCID
classification cs.CV
keywords smalldataimageclassificationaugmentationfeaturemixingsegmentationmasksgenerativeadversarialnetworksCLIPSynsetEntropymulti-imagewithoutpretraining
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

HydraMix is a data-augmentation method that generates new training images by combining an arbitrary number of same-class images in a learned feature space, guided by masks derived from image segmentation. The paper claims that this multi-image feature mixing, trained from scratch with reconstruction and adversarial losses, sets a new state of the art in small-data image classification, outperforming MixUp, GLICO, ChimeraMix, and other baselines on ciFAIR-10, STL-10, and ciFAIR-100 with 5 to 100 samples per class. If this holds, it offers a path to training classifiers without pretraining or large annotated collections, which matters for applications constrained by privacy and copyright. The paper also introduces a CLIP-based text-image metric, CLIP Synset Entropy, to quantify how much class-related variety the augmented images cover.

What carries the argument

The load-bearing component is the mixing module: a segmentation-based mask generator that samples segment regions from each of $N$ images and produces a discrete one-hot mask $M \in [0,1]^{N \times H' \times W'}$ with $\sum_{i=1}^{N} M_{i,h,w} = 1$, so each spatial location of the fused feature map comes from exactly one image. This mask is applied in the feature space of a CycleGAN-style encoder-decoder, trained with reconstruction, Laplacian pyramid perceptual, and adversarial losses. The mask gives the generator semantic guidance about which regions to borrow from which image, while feature-space operation allows blends that pixel-space interpolation cannot produce.

What would settle it

Train a strong classifier on the full (unsubsampled) dataset and measure the fraction of HydraMix-generated images whose predicted label matches the class they were mixed from; if a substantial fraction are mislabeled, the claimed accuracy gains would not reflect genuine class-preserving augmentation. A sharper control is to compare HydraMix against a version that mixes images from different classes: if accuracy does not drop sharply, the benefit does not come from class-guided mixing.

Watch

Extended reading notes

Core claim

The central discovery claimed is that mixing the features of several images from the same class, not just two and not in pixel space, produces synthetic training images whose diversity improves downstream classification in the small-data regime. HydraMix's generator encodes $N$ images, fuses their feature maps according to a discrete segmentation-based mask $M$ via $F_{\mathrm{mix}} = \sum_{i=1}^{N} F_i \odot M_i$, and decodes the result; it is trained with a reconstruction loss, a Laplacian pyramid perceptual loss, and an adversarial loss. On benchmarks with 5 to 100 examples per class, the paper reports consistent gains over prior methods, with the largest margins at the smallest sample sizes, and shows the approach transfers to object detection and can be combined with automatic augmentation.

Load-bearing premise

The pipeline rests on the assumption that a generator trained with only a few dozen images, using reconstruction and adversarial losses, produces new images that stay faithful to the class label while adding useful variety, yet the paper never measures label preservation directly.

Editorial extensions

If this is right

  • On ciFAIR-10, STL-10, and ciFAIR-100, HydraMix reports higher test accuracy than every compared baseline at each sample count from 5 to 100 per class, with the largest margins at 5 samples.
  • HydraMix combines well with automatic augmentation policies, beating AutoAugment and TrivialAugment alone and improving the combined pipeline most at small sample counts.
  • As a proof-of-concept, HydraMix transfers to object detection: on COCO with 1-shot and 3-shot training, it raises Faster R-CNN AP by 28.9% and 19.8%, respectively.
  • HydraMix also strengthens a self-supervised classifier: DINO plus HydraMix reaches 33.20% on STL-10 with 5 samples per class, versus 30.81% for DINO alone.
  • The CLIP Synset Entropy measurements indicate that HydraMix-generated images cover more hyponym concepts per class than the original data or MixUp-generated data.

Reading between the lines

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

  • Beyond the paper: because label preservation is never directly measured, the key open test is whether HydraMix's gains persist when generated images are filtered by a full-data classifier for semantic correctness.
  • Beyond the paper: the CLIP Synset Entropy metric could serve as a general, annotation-free audit of augmentation diversity for other pipelines, not just HydraMix.
  • Beyond the paper: the authors' own suggestion to replace the fixed segmentation with end-to-end unsupervised segmentation would make the pipeline fully unsupervised in its mask generation, potentially extending it to data without any human annotation.
  • Beyond the paper: the cross-domain results hint that a generator trained on one dataset can be reused on another; if reproduced, augmentation generators could be trained once on large unlabeled data and shipped for small-data tasks.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper introduces HydraMix, a generative data-augmentation method that combines an arbitrary number of same-class images in feature space using a segmentation-based mixing mask. The generator is trained with reconstruction, perceptual, and adversarial losses, and the resulting synthetic images are injected into the classifier training pipeline. The method is evaluated on ciFAIR-10, STL-10, ciFAIR-100, and ImageNet (5-shot), with a reported new state of the art in the small-data regime. The paper also proposes a CLIP Synset Entropy metric to assess the diversity of generated images and provides extensive ablations covering the number of mixed images, mixing ratio, generator image size, segmentation noise, and cross-domain transfer.

Significance. If the reported empirical results are statistically reliable, HydraMix is a practical augmentation pipeline for small-data image classification that works without pretraining. The paper strengthens the case with a large set of ablations, a cross-domain analysis, and a proof-of-concept object-detection experiment, and the proposed CLIP Synset Entropy metric is a reasonable attempt to quantify the semantic coverage of generated data. However, the central state-of-the-art claim is currently undermined by the lack of significance testing and by the apparent use of test-set performance to select key hyperparameters. The underlying method is coherent and the ablations support the role of the generator, so the issues appear fixable within a revision.

major comments (3)
  1. [Table 1, Section 4.2] The claim that HydraMix 'outperforms existing state-of-the-art methods' is not supported by the reported statistics. Several of the decisive margins are smaller than the reported standard deviations, for example ciFAIR-10@5 (39.05±2.77 vs 37.31±2.57), STL-10@5 (33.09±1.59 vs 32.18±0.90), and STL-10@100 (60.87±1.55 vs 60.44±0.71). No paired significance test across the five seeds is reported, so it is impossible to tell whether these differences are systematic. The authors should add such tests (e.g., a Wilcoxon signed-rank test or paired bootstrap) or appropriately temper the abstract and conclusion.
  2. [Sections 4.8.1 and 4.8.3] The grid search over the number of mixed images N (Section 4.8.1, Table 7) and the analysis of generator image size (Section 4.8.3, Figure 6) evaluate 'downstream classification performance' without specifying any validation split. The main results in Table 1 use N=4 for all datasets, but Table 7 itself shows that the optimal N varies by dataset and sample count (e.g., STL-10 often favors N=5). If the test set informed the choice of N and generator image size, the reported accuracies are optimistically biased. The paper must describe a validation-based selection procedure, or explicitly acknowledge that these hyperparameters were chosen on the test set and then treat the main results as exploratory.
  3. [Table 3, Section 4.2] The ImageNet 5-shot result in Table 3 shows HydraMix at 9.48 Top-1 versus ChimeraMix at 9.44, a difference of only 0.04 percentage points, yet the text states that both methods 'significantly outperform' the baselines. With no error bars or significance test, this claim is unsupported; the result should be described as parity with ChimeraMix on this benchmark.
minor comments (5)
  1. [Section 4.2] The sentence 'On ciFAIR-100 with 5 images per class, for example, the baseline achieves an accuracy of 31.37%, MixUp of 33.41%, ChimeraMix+Seg of 37.31%, and HydraMix of 39.05%' uses the ciFAIR-10@5 values from Table 1; the corresponding ciFAIR-100@5 numbers are 18.78, 20.63, 21.09, and 24.86. Please correct the text.
  2. [Figure 4 caption] The caption contains a typo: 'with5 samples per class' should be 'with 5 samples per class'.
  3. [Section 4.8.2, Figure 5] Figure 5 reports 'average validation accuracy' but does not show error bars or the number of seeds; adding these would make the sensitivity analysis more informative.
  4. [Section 4.8.4, Figure 8] Figure 8 shows the effect of segmentation noise without error bars, which is particularly important because the reported effect is small relative to the accuracy scale; please add standard deviations.
  5. [Section 4.8.5] The sentence 'On ciFAIR-10 and ciFAIR-100, the best results is achieved when the generator is trained on the same dataset' contains a grammatical error ('results is' should be 'result is' or 'results are').

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central claim is supported by an external downstream classification benchmark and an independently pretrained CLIP metric, with no fitted quantity renamed as a prediction.

full rationale

HydraMix contains no derivation chain whose output is assumed as input. The method generates images via an encoder-mixing-decoder architecture optimized with reconstruction, perceptual, and adversarial losses, and the claimed improvement is measured on held-out test sets of ciFAIR-10, STL-10, ciFAIR-100, ImageNet, and COCO detection. The CLIP Synset Entropy metric uses a pretrained CLIP ViT-B/16 model (LAION-2B via OpenCLIP) that was not involved in training HydraMix, so the diversity claim is an independent measurement rather than a self-defined quantity. The only self-citations are to ChimeraMix (Reinders et al., 2022), which is used as a baseline and as motivation for mixing in feature space; the latter is independently corroborated by the paper's own ablation in Table 6/Figure 4 comparing generator-based mixing with pixel-space mixing. The grid search over the number of mixed images and the mixing ratio is a hyperparameter selection issue, not a circular one: the reported accuracies still come from an external downstream classifier on the test sets, and no parameter is fitted to the final reported quantity in a way that makes the comparison true by construction. Concerns about statistical significance of small margins or test-set-informed hyperparameter choice are correctness/robustness issues, not circularity. Therefore the appropriate circularity score is 0.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim rests on tuned hyperparameters (N, pgen, loss weights, and optionally generator size) and on domain assumptions about label preservation, generator generalization, segmentation usefulness, and the validity of the CLIP-based diversity metric. No new physical or mathematical entities are introduced.

free parameters (5)
  • Number of mixed images N = 4 (grid-searched over 2-5)
    Main results fix N=4 after a grid search in Section 4.8.1; the optimal value varies by dataset and sample count.
  • Mixing probability pgen = 0.5 (optimal range 0.4-0.9 in Figure 5)
    Controls the fraction of generated images used during classifier training; selected empirically in Section 4.8.2.
  • Loss weights alpha_rec, alpha_per, alpha_G,disc = 1000, 1, 1
    Set in Section 4.10 to balance the small magnitude of the MSE reconstruction loss; not derived from first principles.
  • Generator image size = 96x96 for ciFAIR-10/100 in the size analysis
    Section 4.8.3 shows that upscaling the generator input improves small-image datasets; the main experimental configuration does not clearly state whether this setting was used.
  • CLIP Synset Entropy temperature tau = 1/100
    Used only for the proposed diversity metric in Sections 4.5 and 4.10, so it does not affect classification results.
assumptions (4)
  • domain assumption Generated images inherit the class label of the input images because mixing is restricted to same-class images.
    HydraMix generates images by mixing only images from the same class and feeds them to the classifier as labeled samples (Sections 3.2 and 4.1). Label preservation is never directly measured.
  • domain assumption A generator trained with reconstruction and adversarial losses on a handful of examples generalizes well enough to produce useful novel samples.
    All downstream gains depend on this empirical generalization; the paper provides indirect evidence through classification accuracy but no direct proof.
  • domain assumption Felzenszwalb segmentation produces semantically meaningful regions for mixing.
    Segmentation masks are computed offline and drive mask sampling (Section 3.2); robustness to synthetic noise is tested in Section 4.8.4 but true segmentation quality is not varied.
  • domain assumption CLIP similarity to WordNet hyponyms reflects the generality or diversity of generated data.
    The CLIP Synset Entropy metric in Sections 4.5 and 4.9 assumes that higher hyponym coverage measured by CLIP implies better augmentation diversity.

how reviews work

0 comments
Cite this review

Pith. "Pith review of HydraMix: Multi-Image Feature Mixing for Small Data Image Classification." pith.science (2026). https://pith.science/paper/XXSDQE3D

@misc{pith2026250109504,
  author       = {Pith},
  title        = {Pith review of: HydraMix: Multi-Image Feature Mixing for Small Data Image Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XXSDQE3D}},
  note         = {Machine review of arXiv:2501.09504}
}
read the original abstract

Training deep neural networks requires datasets with a large number of annotated examples. The collection and annotation of these datasets is not only extremely expensive but also faces legal and privacy problems. These factors are a significant limitation for many real-world applications. To address this, we introduce HydraMix, a novel architecture that generates new image compositions by mixing multiple different images from the same class. HydraMix learns the fusion of the content of various images guided by a segmentation-based mixing mask in feature space and is optimized via a combination of unsupervised and adversarial training. Our data augmentation scheme allows the creation of models trained from scratch on very small datasets. We conduct extensive experiments on ciFAIR-10, STL-10, and ciFAIR-100. Additionally, we introduce a novel text-image metric to assess the generality of the augmented datasets. Our results show that HydraMix outperforms existing state-of-the-art methods for image classification on small datasets.

Figures

Figures reproduced from arXiv: 2501.09504 by the authors.

Figure 1
Figure 1. HydraMix introduces a novel feature-mixing architecture that combines the content of an arbitrary number of [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Qualitative samples using the HydraMix method [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. CLIP Synset Entropy (↑) of the original dataset, a dataset generated with MixUp, and a dataset generated with HydraMix. By sampling new compositions, HydraMix is able to generate a larger variety of images that cover more synset concepts. examples are shown in [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Average classification accuracy on the evaluated datasets with [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Evaluation of the mixing probability pgen on different datasets. The default mixing ratio of 0.5 is indicated by a dashed line. Experiments are performed with 5 examples per class. 4.6. Generator Impact We investigate the impact of the proposed generator on the generat…
Figure 6
Figure 6. Figure 6: Analysis of the generator image size for different [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 8
Figure 8. Figure 8: Analysis of the impact of noisy segmentation [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 7
Figure 7. Figure 7: Distribution of the synset similarity with respect [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 9
Figure 9. Figure 9: Visualization of the image embeddings with CLIP. The original images (green cross), MixUp images (yellow [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

54 extracted references · 48 canonical work pages

  1. [1]

    S., Li, Z., Salakhutdinov, R., Wang, R., and Yu, D

    Arora, S., Du, S. S., Li, Z., Salakhutdinov, R., Wang, R., and Yu, D. Harnessing the power of infinitely wide deep nets on small-data tasks. In Proc. of ICLR , 2020

    work page 2020

  2. [2]

    and Weinshall, D

    Azuri, I. and Weinshall, D. Generative latent implicit conditional optimization when learning from small sample. In 25th International Conference on Pattern Recognition, ICPR , pp.\ 8584--8591, 2021. doi:10/gn3cdn

    work page 2021

  3. [3]

    and Denzler, J

    Barz, B. and Denzler, J. Deep Learning on Small Datasets without Pre-Training using Cosine Loss . In IEEE Winter Conference on Applications of Computer Vision , WACV , 2020 a . ISBN 978-1-72816-553-0. doi:10/gm84cd

    work page 2020

  4. [4]

    and Denzler, J

    Barz, B. and Denzler, J. Do we train on test data? Purging CIFAR of near-duplicates. Journal of Imaging, 6, 2020 b . doi:10/gjqb98

    work page 2020

  5. [5]

    A., Adeli, E., Altman, R., Arora, S., von Arx , S., Bernstein, M

    Bommasani, R., Hudson, D. A., Adeli, E., Altman, R., Arora, S., von Arx , S., Bernstein, M. S., Bohg, J., Bosselut, A., Brunskill, E., Brynjolfsson, E., Buch, S., Card, D., Castellon, R., Chatterji, N., Chen, A., Creel, K., Davis, J. Q., Demszky, D., Donahue, C., Doumbouya, M., Durmus, E., Ermon, S., Etchemendy, J., Ethayarajh, K., Fei-Fei , L., Finn, C.,...

    work page 2021

  6. [6]

    Small data, big decisions: Model selection in the small-data regime

    Bornschein, J., Visin, F., and Osindero, S. Small data, big decisions: Model selection in the small-data regime. In Proc. of ICML , 2020

    work page 2020

  7. [7]

    Random Forests

    Breiman, L. Random Forests . Machine Learning, 45 0 (1): 0 5--32, October 2001. ISSN 1573-0565. doi:10.1023/A:1010933404324

    doi:10.1023/a:1010933404324 2001

  8. [8]

    and Iocchi, L

    Brigato, L. and Iocchi, L. A close look at deep learning with small data. 25th International Conference on Pattern Recognition, ICPR, pp.\ 2490--2497, 2021. doi:10/gn3cds

    work page 2021

Show all 54 references
  1. [9]

    Tune it or don't use it: Benchmarking data-efficient image classification

    Brigato, L., Barz, B., Iocchi, L., and Denzler, J. Tune it or don't use it: Benchmarking data-efficient image classification. IEEE/CVF International Conference on Computer Vision Workshops, ICCVW, pp.\ 1071--1080, 2021. doi:10/gn3cfv

    2021

  2. [10]

    Brooks, T., Holynski, A., and Efros, A. A. Instructpix2pix: Learning to follow image editing instructions. In Proceedings of the IEEE / CVF Conference on Computer Vision and Pattern Recognition , pp.\ 18392--18402, 2023

    2023

  3. [11]

    Emerging Properties in Self-Supervised Vision Transformers

    Caron, M., Touvron, H., Misra, I., Jégou, H., Mairal, J., Bojanowski, P., and Joulin, A. Emerging Properties in Self-Supervised Vision Transformers . In Proceedings of the International Conference on Computer Vision ( ICCV ) , 2021

    2021

  4. [12]

    An analysis of single-layer networks in unsupervised feature learning

    Coates, A., Ng, A., and Lee, H. An analysis of single-layer networks in unsupervised feature learning. In Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics , volume 15 of Proceedings of Machine Learning Research, pp.\ 215--223. PM...

    2011

  5. [13]

    D., Zoph, B., Man \'e , D., Vasudevan, V., and Le, Q

    Cubuk, E. D., Zoph, B., Man \'e , D., Vasudevan, V., and Le, Q. V. AutoAugment : Learning augmentation strategies from data. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR , pp.\ 113--123, 2019. doi:10/ggvxs4

    2019

  6. [14]

    Dabouei, A., Soleymani, S., Taherkhani, F., and Nasrabadi, N. M. SuperMix : Supervising the mixing data augmentation. IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR, pp.\ 13789--13798, 2021. doi:10/gnqj42

    2021

  7. [15]

    L., Chintala, S., Szlam, A., and Fergus, R

    Denton, E. L., Chintala, S., Szlam, A., and Fergus, R. Deep generative image models using a laplacian pyramid of adversarial networks. In Cortes, C., Lawrence, N. D., Lee, D. D., Sugiyama, M., and Garnett, R. (eds.), Advances in Neural Information Processing Systems 28, pp.\ 1...

    2015

  8. [16]

    and Taylor, G

    DeVries, T. and Taylor, G. W. Improved Regularization of Convolutional Neural Networks with Cutout . ArXiv preprint, abs/1708.04552, 2017

    2017 arXiv

  9. [17]

    An Image is Worth 16x16 Words : Transformers for Image Recognition at Scale

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., Uszkoreit, J., and Houlsby, N. An Image is Worth 16x16 Words : Transformers for Image Recognition at Scale . In ICLR , pp.\ 21, 2021

    2021

  10. [18]

    P., and Posner, I

    Engelcke, M., Jones, O. P., and Posner, I. GENESIS-V2 : Inferring Unordered Object Representations without Iterative Refinement . In Neural Information Processing Systems , 2021

    2021

  11. [19]

    Felzenszwalb, P. F. and Huttenlocher, D. P. Efficient Graph-Based Image Segmentation . International Journal of Computer Vision, 59 0 (2), 2004. ISSN 0920-5691. doi:10/fdmw8q

    2004

  12. [20]

    Parametric Scattering Networks

    Gauthier, S., Th \'e rien, B., Als \`e ne-Racicot , L., Rish, I., Belilovsky, E., Eickenberg, M., and Wolf, G. Parametric Scattering Networks . ArXiv preprint, abs/2107.09539, 2021

    2021 arXiv

  13. [21]

    Deep Residual Learning for Image Recognition

    He, K., Zhang, X., Ren, S., and Sun, J. Deep Residual Learning for Image Recognition . In 2016 IEEE Conference on Computer Vision and Pattern Recognition ( CVPR ) , pp.\ 770--778, Las Vegas, NV, USA , June 2016. IEEE . ISBN 978-1-4673-8851-1. doi:10.1109/CVPR.2016.90

    2016 doi

  14. [22]

    A., and Liang, P

    Henderson, P., Li, X., Jurafsky, D., Hashimoto, T., Lemley, M. A., and Liang, P. Foundation Models and Fair Use , March 2023

    2023

  15. [23]

    OpenCLIP

    Ilharco, G., Wortsman, M., Wightman, R., Gordon, C., Carlini, N., Taori, R., Dave, A., Shankar, V., Namkoong, H., Miller, J., Hajishirzi, H., Farhadi, A., and Schmidt, L. OpenCLIP . Zenodo, July 2021

    2021

  16. [24]

    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

  17. [25]

    SmoothMix : Training Confidence-calibrated Smoothed Classifiers for Certified Robustness

    Jeong, J., Park, S., Kim, M., Lee, H.-C., Kim, D., and Shin, J. SmoothMix : Training Confidence-calibrated Smoothed Classifiers for Certified Robustness . In Advances in Neural Information Processing Systems , November 2021

    2021

  18. [26]

    and Kim, S

    Kang, M. and Kim, S. GuidedMixup : An Efficient Mixup Strategy Guided by Saliency Maps . Proceedings of the AAAI Conference on Artificial Intelligence, 37 0 (1): 0 1096--1104, June 2023. ISSN 2374-3468, 2159-5399. doi:10.1609/aaai.v37i1.25191

    2023 doi

  19. [27]

    Kim, J.-H., Choo, W., and Song, H. O. Puzzle Mix : Exploiting Saliency and Local Statistics for Optimal Mixup . In International Conference on Machine Learning ( ICML ) , December 2020

    2020

  20. [28]

    Krizhevsky, A., Sutskever, I., and Hinton, G. E. ImageNet classification with deep convolutional neural networks. In Bartlett, P. L., Pereira, F. C. N., Burges, C. J. C., Bottou, L., and Weinberger, K. Q. (eds.), Advances in Neural Information Processing Systems 25, pp.\ 1106-...

    2012

  21. [29]

    Li, S., Wang, Z., Liu, Z., Wu, D., Tan, C., Jin, W., and Li, S. Z. OpenMixup : A Comprehensive Mixup Benchmark for Visual Classification , October 2023

    2023

  22. [30]

    MiAMix : Enhancing Image Classification through a Multi-stage Augmented Mixed Sample Data Augmentation Method

    Liang, W., Liang, Y., and Jia, J. MiAMix : Enhancing Image Classification through a Multi-stage Augmented Mixed Sample Data Augmentation Method . Processes, 11 0 (12), August 2023

    2023

  23. [31]

    J., Hays, J., Perona, P., Ramanan, D., Dollár, P., and Zitnick, C

    Lin, T.-Y., Maire, M., Belongie, S. J., Hays, J., Perona, P., Ramanan, D., Dollár, P., and Zitnick, C. L. Microsoft COCO : Common objects in context. In European Conference on Computer Vision, 2014

    2014

  24. [32]

    P., Rademaker, A., Rudnicka, E., and Bond, F

    McCrae, J. P., Rademaker, A., Rudnicka, E., and Bond, F. English WordNet 2020: Improving and extending a WordNet for English using an open-source methodology. In Declerk, T., Gonzalez-Dios , I., and Rigau, G. (eds.), Proceedings of the LREC 2020 Workshop on Multimodal Wordnets...

    2020

  25. [33]

    Miller, G. A. WordNet : A lexical database for English . Communications of the ACM, 38 0 (11): 0 39--41, November 1995. ISSN 0001-0782, 1557-7317. doi:10.1145/219717.219748

    1995

  26. [34]

    and Hutter, F

    M \"u ller, S. and Hutter, F. TrivialAugment : Tuning-free yet state-of-the-art data augmentation. In IEEE / CVF International Conference on Computer Vision ( ICCV ) , 2021

    2021

  27. [35]

    What is being transferred in transfer learning? In Larochelle, H., Ranzato, M., Hadsell, R., Balcan, M

    Neyshabur, B., Sedghi, H., and Zhang, C. What is being transferred in transfer learning? In Larochelle, H., Ranzato, M., Hadsell, R., Balcan, M. F., and Lin, H. (eds.), Advances in Neural Information Processing Systems 33, pp.\ 512--523, 2020

    2020

  28. [36]

    W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., Krueger, G., and Sutskever, I

    Radford, A., Kim, J. W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., Krueger, G., and Sutskever, I. Learning Transferable Visual Models From Natural Language Supervision . In Proceedings of the 38th International Conference on...

    2021

  29. [37]

    Y., and Rosenhahn, B

    Reinders, C., Ackermann, H., Yang, M. Y., and Rosenhahn, B. Object Recognition from very few Training Examples for Enhancing Bicycle Maps . In 2018 IEEE Intelligent Vehicles Symposium ( IV ) , May 2018

    2018

  30. [38]

    ChimeraMix : Image classification on small datasets via masked feature mixing

    Reinders, C., Schubert, F., and Rosenhahn, B. ChimeraMix : Image classification on small datasets via masked feature mixing. In 31st International Joint Conference on Artificial Intelligence and the 25th European Conference on Artificial Intelligence ( IJCAI-ECAI ) , 2022

    2022

  31. [39]

    Faster R-CNN : Towards real-time object detection with region proposal networks

    Ren, S., He, K., Girshick, R., and Sun, J. Faster R-CNN : Towards real-time object detection with region proposal networks. In Cortes, C., Lawrence, N., Lee, D., Sugiyama, M., and Garnett, R. (eds.), Advances in Neural Information Processing Systems, volume 28. Curran Associat...

    2015

  32. [40]

    High- Resolution Image Synthesis with Latent Diffusion Models

    Rombach, R., Blattmann, A., Lorenz, D., Esser, P., and Ommer, B. High- Resolution Image Synthesis with Latent Diffusion Models . In Proceedings of the IEEE / CVF Conference on Computer Vision and Pattern Recognition , April 2022

    2022

  33. [41]

    C., and Fei-Fei , L

    Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., Berg, A. C., and Fei-Fei , L. ImageNet Large Scale Visual Recognition Challenge . International Journal of Computer Vision, 115 0 (3), 2015. ISSN 0920-5691,...

    2015

  34. [42]

    LAION-5B : An open large-scale dataset for training next generation image-text models

    Schuhmann, C., Beaumont, R., Vencu, R., Gordon, C., Wightman, R., Cherti, M., Coombes, T., Katta, A., Mullis, C., Wortsman, M., Schramowski, P., Kundurthy, S., Crowson, K., Schmidt, L., Kaczmarczyk, R., and Jitsev, J. LAION-5B : An open large-scale dataset for training next ge...

    2022

  35. [43]

    The Cost of Training NLP Models : A Concise Overview , April 2020

    Sharir, O., Peleg, B., and Shoham, Y. The Cost of Training NLP Models : A Concise Overview , April 2020

    2020

  36. [44]

    and Zisserman, A

    Simonyan, K. and Zisserman, A. Very deep convolutional networks for large-scale image recognition. In Bengio, Y. and LeCun, Y. (eds.), Proc. of ICLR , 2015

    2015

  37. [45]

    Manifold Mixup : Better Representations by Interpolating Hidden States

    Verma, V., Lamb, A., Beckham, C., Najafi, A., Mitliagkas, I., Lopez-Paz , D., and Bengio, Y. Manifold Mixup : Better Representations by Interpolating Hidden States . In ICML , 2019

    2019

  38. [46]

    and Herley, C

    Vetterli, M. and Herley, C. Wavelets and Filter Banks : Theory and Design . IEEE Transactions on Signal Processing, 40 0 (9): 0 2207--2232, October 1992. doi:10.1109/78.157221

    1992 doi

  39. [47]

    Understanding How Dimension Reduction Tools Work : An Empirical Approach to Deciphering t- SNE , UMAP , TriMap , and PaCMAP for Data Visualization

    Wang, Y., Huang, H., Rudin, C., and Shaposhnik, Y. Understanding How Dimension Reduction Tools Work : An Empirical Approach to Deciphering t- SNE , UMAP , TriMap , and PaCMAP for Data Visualization . Journal of Machine Learning Research, 22 0 (201): 0 1--73, 2021

    2021

  40. [48]

    Towards Fairer Datasets : Filtering and Balancing the Distribution of the People Subtree in the ImageNet Hierarchy

    Yang, K., Qinami, K., Fei-Fei , L., Deng, J., and Russakovsky, O. Towards Fairer Datasets : Filtering and Balancing the Distribution of the People Subtree in the ImageNet Hierarchy . In Proceedings of the 2020 Conference on Fairness , Accountability , and Transparency , pp.\ 5...

    2020

  41. [49]

    J., Yoo, Y., and Choe, J

    Yun, S., Han, D., Chun, S., Oh, S. J., Yoo, Y., and Choe, J. CutMix : Regularization strategy to train strong classifiers with localizable features. In IEEE / CVF International Conference on Computer Vision, ICCV , pp.\ 6022--6031, 2019. doi:10/ghfg25

    2019

  42. [50]

    and Komodakis, N

    Zagoruyko, S. and Komodakis, N. Wide Residual Networks . In Proceedings of the British Machine Vision Conference 2016, BMVC , 2016

    2016

  43. [51]

    N., and Lopez-Paz , D

    Zhang, H., Ciss \'e , M., Dauphin, Y. N., and Lopez-Paz , D. Mixup: Beyond empirical risk minimization. In Proc. of ICLR , 2018

    2018

  44. [52]

    Random erasing data augmentation

    Zhong, Z., Zheng, L., Kang, G., Li, S., and Yang, Y. Random erasing data augmentation. In The Thirty-Fourth AAAI Conference on Artificial Intelligence, AAAI , pp.\ 13001--13008, 2020

    2020

  45. [53]

    Zhu, J.-Y., Park, T., Isola, P., and Efros, A. A. Unpaired Image-to-Image Translation Using Cycle-Consistent Adversarial Networks . In IEEE International Conference on Computer Vision , ICCV , pp.\ 2242--2251, Venice , October 2017. IEEE . ISBN 978-1-5386-1032-9. doi:10/gfhw33

    2017

  46. [54]

    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 10, 2026 · model on record in the stance chip above.