Pith. sign in

REVIEW 3 major objections 5 minor 29 references

DAM: Domain-Aware Module for Multi-Domain Dataset Condensation

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read A plug-in domain-aware module improves dataset condensation by embedding spatial domain masks into synthetic images, boosting in-domain, out-of-domain, and cross-architecture accuracy without adding images per class.

desk verdict Introduces a useful new task and a plausible module, but the MTT evidence is confounded by an unfair baseline; the core idea is still worth a serious look. read the letter →

arxiv 2505.22387 v1 pith:UMWNBHRY submitted 2025-05-28 cs.CV cs.AIcs.LG

Jaehyun Choi , Gyojin Han , Dong-Jae Lee , Sunghyun Baek , Junmo Kim This is my paper
classification cs.CVcs.AIcs.LG
keywords datasetcondensationmulti-domaindomain-awaremodulepseudo-domainlabelingfrequency-basedspatialmasksdomaingeneralizationcross-architecture
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

Dataset condensation squeezes a large training set into a tiny synthetic one, but existing methods assume the source imagery shares a single visual style. This paper argues that real-world datasets are multi-domain, with photos, cartoons, sketches, and paintings mixed together, and that ignoring this causes condensed data to collapse toward dominant styles and lose accuracy. It introduces Multi-Domain Dataset Condensation (MDDC) and the Domain-Aware Module (DAM), a plug-in training-time component that superimposes learnable spatial masks onto each synthetic image so different domains occupy different regions. Because explicit domain labels are usually unavailable, DAM derives pseudo-domain labels from the mean amplitude of low-frequency Fourier components of the real images. The claim is that this consistently improves in-domain, out-of-domain, and cross-architecture accuracy across DC, DM, and MTT baselines without increasing images per class.

What carries the argument

The load-bearing object is the spatial domain mask $\alpha^{d,i}_m \in \mathbb{R}^{H \times W \times 3}$, computed as a per-pixel temperature softmax over D learnable mask tensors $z^{d,i}_m$. Multiplying the synthetic image $\tilde{x}^i_m$ element-wise by each mask gives domain-specific views $\tilde{x}^{d,i}_m$, and the identity $\tilde{x}^i_m = \sum_{d=0}^{D-1} \tilde{x}^{d,i}_m$ holds because the softmax outputs sum to 1, so no information is lost when the image is split into domain views. Domain supervision comes from frequency-based pseudo-domain labels: each real image is assigned one of D bins by ranking its mean low-frequency FFT amplitude, a heuristic borrowed from Fourier-based domain adaptation and generalization. The domain branch of the loss uses the same base condensation loss (gradient matching, distribution matching, or trajectory matching) but grouped by pseudo-domain instead of by class. This machinery lets domain structure be embedded during training while leaving the final synthetic dataset and its IPC unchanged.

What would settle it

Train DAM on a purpose-built multi-domain dataset where the domain shift is deliberately placed in high-frequency content (e.g., texture swaps) so that mean low-frequency amplitude is nearly constant across domains; if DAM does not beat the base condensation method there, the FFT pseudo-labeling assumption is the failing component.

Watch

Extended reading notes

Core claim

The paper's central claim is that domain diversity can be encoded inside a synthetic dataset without extra images or labels. DAM attaches D learnable spatial masks to each synthetic image, normalizes them with a per-pixel softmax so they sum to one, and multiplies each mask against the image to produce D domain-specific views that reconstruct the original image exactly. A second branch of the condensation loss trains these masks against pseudo-domain labels, obtained by sorting real images by mean low-frequency FFT amplitude and slicing the sorted list into D bins. During condensation the synthetic image is updated by both class and domain losses, but after condensation the masks are discarded, so downstream models see only the original unmodified synthetic images. The reported result is consistent improvement over three prior methods on five datasets, with the largest gains on multi-domain benchmarks such as PACS where 1-IPC MTT accuracy rises from 39.7 to 46.6.

Load-bearing premise

The load-bearing assumption is that a single low-frequency image statistic can split a mixed dataset into pseudo-domains that capture the visual variation that matters for classification.

Editorial extensions

If this is right

  • DAM can be plugged into gradient-matching, distribution-matching, and trajectory-matching condensation methods without changing the number of synthetic images or their class budget.
  • Condensed data becomes more transferable: accuracy on unseen domains improves, as in leave-one-domain-out tests on PACS, VLCS, and Office-Home.
  • Cross-architecture generalization improves, including to ViT-Tiny and ViT-Small, which the original condensation papers did not evaluate.
  • Multi-domain condensation no longer requires explicit domain labels or per-domain condensation, which would inflate the synthetic dataset size linearly with the number of domains.
  • The performance gap between single-domain and multi-domain condensation narrows, with some target domains matching single-domain performance.

Reading between the lines

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

  • If low-frequency amplitude sorting really captures the domain structure that matters, the same pseudo-labeling could be reused beyond condensation, for example to guide data selection or augmentation schedules in other data-efficient learning pipelines.
  • The exact reconstruction identity means DAM is a form of input-space regularization; one testable extension is whether the learned masks can be reused or transferred to new condensation runs rather than discarded.
  • Because DAM improves out-of-domain generalization, it could serve as a cheap proxy for domain-generalization benchmarking before collecting more data, though the paper only tests this indirectly with leave-one-domain-out evaluation.
  • The optimal number of pseudo-domains D appears dataset-dependent, as the CIFAR-10 sweep shows, suggesting that pseudo-domain granularity itself carries task-relevant information that could be tuned automatically.
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 the Domain-Aware Module (DAM), a training-time module for dataset condensation that equips each synthetic image with D learnable spatial masks, supervised by pseudo-domain labels derived from low-frequency FFT amplitude sorting. The module is plugged into DC, DM, and MTT; the final condensed images themselves remain the same size and IPC. Experiments on CIFAR-10/100, Tiny ImageNet, PACS, VLCS, Office-Home, and DomainNet report in-domain, cross-architecture, and leave-one-domain-out gains, along with ablations of pseudo-label strategies, domain count, and hyperparameters.

Significance. If the results hold, DAM is a simple and useful plug-in for dataset condensation in heterogeneous and multi-domain data, with a plausible mechanism (spatial domain masks) and a practical pseudo-labeling scheme that needs no explicit domain annotations. The paper is strengthened by systematic ablations, error bars over 10 runs, cross-architecture evaluation, and an honest limitation discussion of condensation-time overhead. The main risk is the MTT comparison, which is not controlled; this must be fixed before the general claim of consistent improvement is accepted.

major comments (3)
  1. [Section 4.2, Appendix F, Table H] The MTT comparisons are not controlled, and they are load-bearing for the paper's central claim. Section 4.2 states that 'all of the hyperparameters introduced in each prior method are set identically,' but Appendix F reports that MTT required a separate hyperparameter search when combined with DAM and Gaussian-noise initialization, with chosen values listed in Table H. No evidence is provided that the vanilla MTT baseline was re-tuned under the same Gaussian-noise initialization or the same search budget. This matters because Table A shows that real-image initialization gives vanilla MTT 65.3 on CIFAR-10 at 10 IPC, versus 50.7 in Table 1 under Gaussian initialization; under real initialization, MTT+DAM improves by only +0.2 on CIFAR-10 10 IPC and +0.1 on PACS 1 IPC. The headline MTT+DAM gains (e.g., PACS 1 IPC 39.7 to 46.6, CIFAR-10 10 IPC 50.7 to 57.9) may therefore reflect hyperparameter tuning rather than DAM. Please provide a controlled comparison in which vanilla MTT is re-tuned with Gaussian-noise initialization under the same search budget, or report both initialization settings with matched hyperparameters for all compared methods.
  2. [Section 3.3, Eq. (7)] The domain loss is under-specified. D_dom_syn is defined as {(x̃_d_m, ỹ_m)} with class labels ỹ_m, but the text says that for the domain loss the real batch is grouped by pseudo-domain label. If L_base is the same loss as the class loss (e.g., gradient matching or distribution matching), it needs a consistent label set for both the real and synthetic batches. As written, the reader cannot tell whether the synthetic masked images are labeled with pseudo-domain labels, class labels, or both; the notation suggests class labels, which would make the domain loss inconsistent with the claimed pseudo-domain grouping. Please specify exactly how L_base is adapted to the pseudo-domain task, including the label sets for D_real and D_dom_syn, and how the per-image masks α_d_m are associated with pseudo-domain indices.
  3. [Section 4.2 and Table A] The choice of Gaussian-noise initialization deserves more careful treatment in the main text. The paper justifies the choice by privacy-preserving goals, which is reasonable, but the supplementary results show that the MTT+DAM advantage is much smaller under real-image initialization. Since the main claim is 'consistently improves,' the presentation should make clear that the MTT gains in Tables 1 and 2 depend on the interaction of DAM with Gaussian initialization and the separately tuned hyperparameters, rather than presenting them as unconditional improvements. At minimum, the main text should reference the magnitude of the real-initialization results when the MTT rows are discussed.
minor comments (5)
  1. [Table 6] The checkmark table for pseudo-domain labeling strategies is difficult to parse because the column headers do not clearly separate the feature extractor (FFT, log-Var) from the clustering/ordering strategy (Mean-Sort, K-Means). Please label the columns explicitly and define what each checkmark combination means.
  2. [Supplementary C, Tables B-D] The sentence 'Due to observed instability at 1 IPC, we omit MTT + DAM results for IPC 10 in this ablation' appears in several places, but the omitted rows are for IPC 10 while the stated instability is at 1 IPC. Please clarify the wording so the reason for omitting IPC 10 is unambiguous.
  3. [Figure 4] The caption refers to a dashed red line and a solid blue curve, but the rendered figure appears to be in grayscale; please ensure the legend remains readable in print and that the visual distinction between the baseline and DAM curves is clear.
  4. [Section 5] There is a typo in the sentence about the effect of the number of pseudo-domains: 'even with random number of domains D' should be 'even with a random number of domains D' or 'for any number of domains D.' Also, 'basline' appears once in the same section and should be corrected.
  5. [Section 3.2, Eq. (6)] The pseudo-domain assignment formula uses N/D and integer brackets, but it is not stated how ties are broken or how N is handled when it is not divisible by D; adding a sentence to define the rounding convention would improve reproducibility.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: DAM's gains come from empirical comparisons against external baselines, and the MTT hyperparameter-tuning asymmetry is a comparison-fairness risk, not a definitional loop.

full rationale

The paper's claimed derivation chain is self-contained and contains no definitional loop. DAM adds a domain-aware loss L_dom = L_base(Theta'; D_real, D_dom_syn) (Eq. 7) computed on masked copies x̃_d = x̃ ⊙ α_d, with the per-pixel softmax guaranteeing the exact reconstruction identity x̃ = Σ_d x̃_d (Eq. 3). Because the softmax normalization forces the masks to sum to one pixelwise, the domain decomposition injects no extra information and removes none by construction; whether the auxiliary objective helps is an empirical question, which the paper answers by comparing against the unmodified baselines DC, DM, and MTT using fixed prior-method hyperparameters. The frequency-based pseudo-domain labels are computed from the same real data used for the class loss (same-data auxiliary supervision), not from evaluation labels or the test set; Table 6 shows the specific pseudo-labeling choice (FFT+Mean-Sort) beats random labels and log-variance alternatives, so the domain signal is not a trivial restatement of the method's own outputs. No 'prediction' is fitted and then renamed: the introduced hyperparameters (z_m initialization, λ, τ, D) are ablated in Appendix C and chosen by sweeps, and the reported gains over DC/DM/MTT are not obtained by re-fitting those baselines. The only self-citation, [24] (a prior domain-generalization paper by co-author Jaehyun Choi), appears in a related-work survey sentence and carries no weight in the derivation. Two weaknesses deserve note but are not circularity. First, Appendix F states MTT 'required a separate hyperparameter search due to frequent occurrences of NaN losses' when combined with DAM and Gaussian-noise initialization, with Table H listing tuned MTT+DAM settings; no evidence shows the vanilla MTT baseline was re-tuned under the same Gaussian-noise initialization, so the large MTT gains (e.g., PACS 1 IPC 39.7→46.6) may partly reflect an uneven comparison. This is a comparison-fairness risk, not a definitional reduction, since MTT+DAM is still an independently optimized pipeline evaluated on held-out test data. Second, Appendix B shows real-image initialization shrinks MTT+DAM gains to +0.2 (CIFAR-10, 10 IPC) and +0.1 (PACS, 1 IPC), underscoring that the headline MTT numbers depend on the Gaussian-noise setting; again a validity concern. Neither issue makes the central claim equivalent to its inputs, so the derivation is independent of its own conclusions.

Assumptions & free parameters 6 free parameters · 3 assumptions · 0 invented entities

The central claim rests on a handful of hand-chosen hyperparameters (D, tau, z_m, lambda, beta, sort order) and on the domain assumption that FFT low-frequency amplitude sorting yields meaningful pseudo-domains. No new physical entities are introduced.

free parameters (6)
  • D (number of pseudo-domains) = 4
    Set to 4 for all experiments; ablation in Figure 4 shows best at D=4 for PACS but varying for CIFAR-10. It is a hyperparameter selected on validation data.
  • tau (softmax temperature) = 0.1
    Chosen via sweep (Appendix C.3); lower temperature gives peaked masks. The value 0.1 yields best or second-best across settings.
  • domain mask initialization z_m = 0.01
    Chosen via sweep (Appendix C.1); value selected for best performance and robustness.
  • lambda (domain embedding weight) = 0.1 for DC/DM, 0.01 for MTT
    Chosen via sweep (Appendix C.2); balances class and domain losses. MTT uses smaller weight due to instability.
  • beta (FFT central crop ratio) = not stated
    Required to compute mu_n in Eq. (5); no value is reported in the paper, which hampers reproduction.
  • sort order for pseudo-domain assignment = not stated (ascending or descending)
    Paper says 'ascending or descending order' but does not specify which is used in experiments.
assumptions (3)
  • domain assumption Low-frequency FFT amplitude statistics are a valid proxy for visual domain identity (Section 3.2).
    The pseudo-domain labels are derived solely from the mean amplitude of the low-frequency FFT components; the paper cites prior DA/DG works but provides no direct validation that this correlates with the domains that matter for classification.
  • domain assumption The domain-aware loss computed with pseudo-domain labels and masked synthetic images encourages each image to encode multiple useful domains without degrading class information.
    This is the core mechanism; the paper demonstrates it empirically via ablations, but there is no formal guarantee. The loss definition itself is ambiguous.
  • standard math Standard properties of FFT, softmax normalization, and element-wise multiplication hold as used in Eqs. (1)-(5).
    These are standard mathematical operations; no issue.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DAM: Domain-Aware Module for Multi-Domain Dataset Condensation." pith.science (2026). https://pith.science/paper/UMWNBHRY

@misc{pith2026250522387,
  author       = {Pith},
  title        = {Pith review of: DAM: Domain-Aware Module for Multi-Domain Dataset Condensation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UMWNBHRY}},
  note         = {Machine review of arXiv:2505.22387}
}
read the original abstract

Dataset Condensation (DC) has emerged as a promising solution to mitigate the computational and storage burdens associated with training deep learning models. However, existing DC methods largely overlook the multi-domain nature of modern datasets, which are increasingly composed of heterogeneous images spanning multiple domains. In this paper, we extend DC and introduce Multi-Domain Dataset Condensation (MDDC), which aims to condense data that generalizes across both single-domain and multi-domain settings. To this end, we propose the Domain-Aware Module (DAM), a training-time module that embeds domain-related features into each synthetic image via learnable spatial masks. As explicit domain labels are mostly unavailable in real-world datasets, we employ frequency-based pseudo-domain labeling, which leverages low-frequency amplitude statistics. DAM is only active during the condensation process, thus preserving the same images per class (IPC) with prior methods. Experiments show that DAM consistently improves in-domain, out-of-domain, and cross-architecture performance over baseline dataset condensation methods.

Figures

Figures reproduced from arXiv: 2505.22387 by the authors.

Figure 1
Figure 1. Performance of single- and multi-domain training for existing dataset condensation meth￾ods (DC, DM, MTT) on the PACS dataset under a 10 images per class setting. In the single-domain setup, models are trained using only Cartoon do￾main images, assuming access to explicit domain labels. In contrast, the multi-domain dataset set￾ting trains on the full PACS dataset without do￾main supervision, reflecting modern datas… view at source ↗
Figure 2
Figure 2. DAM incorporates both class-aware training (left) from prior methods and domain-aware [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Visualization of the final output in CIFAR-10 and PACS under 10 IPC setting. The shown [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Experiment with a varying number of domains [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: Visualization of the final output and domain masks in CIFAR-10 under 10 IPC setting. The [PITH_FULL_IMAGE:figures/full_fig_p017_5.png]
Figure 6
Figure 6. Figure 6: Visualization of the final output and domain masks in PACS under 10 IPC setting. The [PITH_FULL_IMAGE:figures/full_fig_p018_6.png]
Figure 7
Figure 7. Figure 7: Visualization of the final output and domain masks in CIFAR-10 under 10 IPC setting. The [PITH_FULL_IMAGE:figures/full_fig_p019_7.png]
Figure 8
Figure 8. Figure 8: Visualization of the final output and domain masks in PACS under 10 IPC setting. The [PITH_FULL_IMAGE:figures/full_fig_p020_8.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

29 extracted references · 22 canonical work pages

  1. [1]

    Dataset distillation,

    T. Wang, J.-Y . Zhu, A. Torralba, and A. A. Efros, “Dataset distillation,”arXiv preprint arXiv:1811.10959, 2018

    arXiv 2018

  2. [2]

    Dataset condensation with gradient matching,

    B. Zhao, K. R. Mopuri, and H. Bilen, “Dataset condensation with gradient matching,” inInternational Conference on Learning Representations, 2021

    work page 2021

  3. [3]

    Dataset condensation with distribution matching,

    B. Zhao and H. Bilen, “Dataset condensation with distribution matching,” inProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2023, pp. 6514–6523

    work page 2023

  4. [4]

    Dataset distillation by matching training trajectories,

    G. Cazenavette, T. Wang, A. Torralba, A. A. Efros, and J.-Y . Zhu, “Dataset distillation by matching training trajectories,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 4750–4759

    work page 2022

  5. [5]

    Learning multiple layers of features from tiny images,

    A. Krizhevsky, G. Hintonet al., “Learning multiple layers of features from tiny images,” 2009

    2009

  6. [6]

    Deeper, broader and artier domain generalization,

    D. Li, Y . Yang, Y .-Z. Song, and T. M. Hospedales, “Deeper, broader and artier domain generalization,” in Proceedings of the IEEE international conference on computer vision, 2017, pp. 5542–5550

    work page 2017

  7. [7]

    Fda: Fourier domain adaptation for semantic segmentation,

    Y . Yang and S. Soatto, “Fda: Fourier domain adaptation for semantic segmentation,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 4085–4095

    work page 2020

  8. [8]

    A fourier-based framework for domain generalization,

    Q. Xu, R. Zhang, Y . Zhang, Y . Wang, and Q. Tian, “A fourier-based framework for domain generalization,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 14 383–14 392

    work page 2021

Show all 29 references
  1. [9]

    Dataset condensation via efficient synthetic-data parameterization,

    J.-H. Kim, J. Kim, S. J. Oh, S. Yun, H. Song, J. Jeong, J.-W. Ha, and H. O. Song, “Dataset condensation via efficient synthetic-data parameterization,” inInternational Conference on Machine Learning. PMLR, 2022, pp. 11 102–11 118

    2022

  2. [10]

    Improved distribution matching for dataset condensation,

    G. Zhao, G. Li, Y . Qin, and Y . Yu, “Improved distribution matching for dataset condensation,” inProceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 7856–7865

    2023

  3. [11]

    Towards lossless dataset distillation via difficulty-aligned trajectory matching,

    Z. Guo, K. Wang, G. Cazenavette, H. LI, K. Zhang, and Y . You, “Towards lossless dataset distillation via difficulty-aligned trajectory matching,” inThe Twelfth International Conference on Learning Representations, 2024. [Online]. Available: https://openreview.net/forum?id=rTBL8OhdhH

    2024

  4. [12]

    Generalizing dataset distillation via deep generative prior,

    G. Cazenavette, T. Wang, A. Torralba, A. A. Efros, and J.-Y . Zhu, “Generalizing dataset distillation via deep generative prior,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 3739–3748

    2023

  5. [13]

    Dˆ 4: Dataset distillation via disentangled diffusion model,

    D. Su, J. Hou, W. Gao, Y . Tian, and B. Tang, “Dˆ 4: Dataset distillation via disentangled diffusion model,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 5809–5818

    2024

  6. [14]

    Accelerating dataset distillation via model augmentation,

    L. Zhang, J. Zhang, B. Lei, S. Mukherjee, X. Pan, B. Zhao, C. Ding, Y . Li, and D. Xu, “Accelerating dataset distillation via model augmentation,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 11 950–11 959

    2023

  7. [15]

    Minimizing the accumulated trajectory error to improve dataset distillation,

    J. Du, Y . Jiang, V . Y . Tan, J. T. Zhou, and H. Li, “Minimizing the accumulated trajectory error to improve dataset distillation,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2023, pp. 3749–3758

    2023

  8. [16]

    Domain-adversarial training of neural networks,

    Y . Ganin, E. Ustinova, H. Ajakan, P. Germain, H. Larochelle, F. Laviolette, M. March, and V . Lempitsky, “Domain-adversarial training of neural networks,”Journal of machine learning research, vol. 17, no. 59, pp. 1–35, 2016

    2016

  9. [17]

    Conditional adversarial domain adaptation,

    M. Long, Z. Cao, J. Wang, and M. I. Jordan, “Conditional adversarial domain adaptation,”Advances in neural information processing systems, vol. 31, 2018

    2018

  10. [18]

    Bridging theory and algorithm for domain adaptation,

    Y . Zhang, T. Liu, M. Long, and M. Jordan, “Bridging theory and algorithm for domain adaptation,” in International conference on machine learning. PMLR, 2019, pp. 7404–7413

    2019

  11. [19]

    Patch-mix transformer for unsupervised domain adaptation: A game perspective,

    J. Zhu, H. Bai, and L. Wang, “Patch-mix transformer for unsupervised domain adaptation: A game perspective,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2023, pp. 3561–3571

    2023

  12. [20]

    Domain generalization via conditional invariant representa- tions,

    Y . Li, M. Gong, X. Tian, T. Liu, and D. Tao, “Domain generalization via conditional invariant representa- tions,” inProceedings of the AAAI conference on artificial intelligence, vol. 32, no. 1, 2018. 10

    2018

  13. [21]

    Domain generalization with mixstyle,

    K. Zhou, Y . Yang, Y . Qiao, and T. Xiang, “Domain generalization with mixstyle,”arXiv preprint arXiv:2104.02008, 2021

    2021 arXiv

  14. [22]

    Swad: Domain generalization by seeking flat minima,

    J. Cha, S. Chun, K. Lee, H.-C. Cho, S. Park, Y . Lee, and S. Park, “Swad: Domain generalization by seeking flat minima,”Advances in Neural Information Processing Systems, vol. 34, pp. 22 405–22 418, 2021

    2021

  15. [23]

    Pcl: Proxy-based contrastive learning for domain generalization,

    X. Yao, Y . Bai, X. Zhang, Y . Zhang, Q. Sun, R. Chen, R. Li, and B. Yu, “Pcl: Proxy-based contrastive learning for domain generalization,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 7097–7107

    2022

  16. [24]

    Tcx: Texture and channel swappings for domain generaliza- tion,

    J. Choi, H. S. Seong, S. Park, and J.-P. Heo, “Tcx: Texture and channel swappings for domain generaliza- tion,”Pattern Recognition Letters, vol. 175, pp. 74–80, 2023

    2023

  17. [25]

    Tiny imagenet visual recognition challenge,

    Y . Le and X. Yang, “Tiny imagenet visual recognition challenge,”CS 231N, vol. 7, no. 7, p. 3, 2015

    2015

  18. [26]

    Unbiased metric learning: On the utilization of multiple datasets and web images for softening bias,

    C. Fang, Y . Xu, and D. N. Rockmore, “Unbiased metric learning: On the utilization of multiple datasets and web images for softening bias,” inProceedings of the IEEE International Conference on Computer Vision, 2013, pp. 1657–1664

    2013

  19. [27]

    Deep hashing network for unsu- pervised domain adaptation,

    H. Venkateswara, J. Eusebio, S. Chakraborty, and S. Panchanathan, “Deep hashing network for unsu- pervised domain adaptation,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017, pp. 5018–5027

    2017

  20. [28]

    Dynamic few-shot visual learning without forgetting,

    S. Gidaris and N. Komodakis, “Dynamic few-shot visual learning without forgetting,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 4367–4375

    2018

  21. [29]

    Moment matching for multi-source domain adaptation,

    X. Peng, Q. Bai, X. Xia, Z. Huang, K. Saenko, and B. Wang, “Moment matching for multi-source domain adaptation,” inProceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 1406–1415. 11 A Details regardingΘandΘ ′ We clarify the roles of two sets of mo...

    2019

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.