Pith. sign in

REVIEW 4 major objections 5 minor 64 references

TIDE: Training Locally Interpretable Domain Generalization Models Enables Test-time Correction

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read TIDE claims that single-source domain generalization is substantially improved by making models localize per-class concepts and by correcting predictions at test time against stored concept signatures.

desk verdict The gains are real if the saliency maps transfer, but that premise is untested. read the letter →

arxiv 2411.16788 v2 pith:5SC7MVP3 submitted 2024-11-25 cs.CV

Aishwarya Agarwal , Srikrishna Karanam , Vineet Gandhi This is my paper · ORCID
classification cs.CV
keywords single-sourcedomaingeneralizationconceptsaliencymapsdiffusionmodelstest-timecorrectioninterpretabilitysignaturescontrastivelearningdomain-invariantfeatures
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

TIDE claims that single-source domain generalization is better served by forcing a model to recognize and localize per-class local concepts than by piling on image augmentations. The paper builds an automatic annotation pipeline that uses a language model to name class-identifying concepts and a diffusion model to draw one example per class with attention maps for those concepts, then transfers those maps to real training images. On top of that, TIDE trains a classifier with two losses: one aligning predicted concept saliency maps with transferred maps, and one contrasting concept-specific features so the same concept clusters across domains. At test time, the predicted concept maps are compared against stored prototypical concept signatures, and the model iteratively masks away regions that do not match until features align, correcting misclassifications. Across PACS, VLCS, OfficeHome, and DomainNet, TIDE reports average gains of 8.33, 13.37, 16.16, and 8.84 points over the second-best approach on each benchmark.

What carries the argument

The load-bearing object is the concept-level saliency map $G^k_x$, a binary mask locating concept $k$ in image $x$, generated once per class from a diffusion model's cross-attention maps and transferred to real images via diffusion feature correspondence. From these maps TIDE defines concept-specific feature vectors $f^k_x$ by pooling convolutional features under the mask, and averages them over training samples to obtain concept signatures $p^k$. The training machinery is a pair of losses: the concept saliency alignment loss $\mathcal{L}_{\mathrm{CSA}}$ forces predicted Grad-CAM maps $S^k_x$ toward $G^k_x$, and the local concept contrastive loss $\mathcal{L}_{\mathrm{LCC}}$ organizes the $f^k_x$ vectors so that the same concept clusters together while different concepts separate. At test time the algorithm compares $f^k_x$ with the stored $p^k$; a cosine distance above threshold $\delta$ triggers an iterative loop that masks the current class saliency from the image and re-predicts until concept features match the signatures or the iteration cap $T$ is reached.

What would settle it

Measure the transferred concept maps against human-annotated concept regions for a few hundred images across PACS and VLCS; if the maps localize concepts like beak or eyes no better than the model's unguided attention, then the alignment loss and concept signatures have no reliable ground truth to operate on.

Watch

Extended reading notes

Core claim

The paper's central claim is that domain shift is best survived by local, class-defining concepts such as beak and feathers for birds or eyes and lips for people, rather than by global features learned from augmentation-heavy training. To test this, TIDE first constructs per-class ground-truth saliency maps automatically: GPT-3.5 lists the distinguishing concepts, a diffusion model generates one exemplar per class with cross-attention maps for those concepts, and diffusion feature transfer maps those maps onto real images in the training set. TIDE then trains with a concept saliency alignment loss that forces the network's Grad-CAM maps to match these transferred maps and with a local concept contrastive loss that pulls concept-specific feature vectors of the same concept together while pushing different concepts apart. At test time, predicted concept features are compared with stored concept-signature prototypes; when they deviate, the model iteratively masks the most salient regions and re-predicts until the concept features align. The paper reports that this scheme establishes a new state of the art on the four standard single-source domain generalization benchmarks.

Load-bearing premise

The method's ground-truth concept maps are produced by transferring saliency regions from one generated example per class to every real training image, and the entire training and test-time correction depend on those maps being accurate; the paper shows visual examples but no quantitative measure of map quality.

Editorial extensions

If this is right

  • On PACS, VLCS, OfficeHome, and DomainNet, TIDE reports average accuracies of 80.02%, 77.08%, 74.01%, and 82.14%, beating the prior best method on each benchmark by 8.33, 13.37, 16.16, and 8.84 points.
  • The local concept contrastive loss produces concept clusters that are compact within a concept and separated across concepts when measured across source and target domains, as shown in the paper's t-SNE visualizations.
  • On the PACS photo-to-sketch split, test-time correction raises accuracy from 74.79% to 82.29%, with correction invoked in 27.8% of samples and converging to the right class in 52.5% of those cases.
  • A TIDE model can be visually interpreted through predicted concept saliency maps, which localize class-defining regions such as beak, eyes, or strings across unseen domains.
  • The full pipeline needs no human annotation: the language model names concepts, the diffusion model draws an exemplar with attention maps, and diffusion feature transfer maps those maps to real images.

Reading between the lines

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

  • An extension the paper leaves implicit is that the test-time correction loop is not tied to TIDE's own training; any classifier that can produce concept saliency maps and has access to concept signatures could in principle run the same masking refinement.
  • The ablation attributes most of the accuracy gain to the contrastive loss, so a plausible reading is that the saliency-alignment loss mainly buys interpretability and the correction mechanism, while the contrastive loss carries the generalization; the paper does not separate these effects.
  • Because the annotation pipeline uses one diffusion exemplar per class, a natural stress test is to vary the exemplar or sample several: if performance is sensitive to that choice, the current results partly reflect a favorable draw rather than concept structure.
  • The failure cases in the appendix show correction can be fooled when image regions resemble concepts of another class, such as an elephant's trunk matching guitar strings, so a confidence-aware version of the signature check is a concrete next step the paper identifies only as future work.
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

4 major / 5 minor

Summary. The paper proposes TIDE, a single-source domain generalization method that trains a classifier to localize class-specific local concepts and uses those localizations at test time to correct predictions. Concept annotations are generated automatically: GPT-3.5 lists per-class concepts, a diffusion model synthesizes one exemplar per class and provides cross-attention maps, and DIFT transfers those maps to real training images. Training combines class and concept cross-entropy losses with a concept saliency alignment loss (Eq. 3) and a local concept contrastive loss (Eq. 5). At inference, an iterative algorithm masks predicted-class saliency regions and compares concept-pooled features against stored concept signatures (Eq. 6, Algorithm 1). The paper reports large accuracy gains over prior SSDG methods on PACS, VLCS, OfficeHome, and DomainNet, with an average improvement of about 12% over the second-best approach, and supports this with ablations and qualitative visualizations.

Significance. If the reported results are trustworthy, TIDE would be a strong new state of the art for single-source domain generalization, with the additional virtue of producing interpretable concept-level saliency maps and a novel test-time correction mechanism. The annotation pipeline is creative and potentially scalable, and the idea of verifying prediction consistency against local concept prototypes is interesting. The paper also provides honest failure-case examples in Appendix A.1. However, the current manuscript does not quantitatively validate the automatically generated concept maps, does not verify the central premise that predicted saliency maps approximate ground-truth maps on unseen domains, and relies on partially unproven baseline numbers for three of the four benchmarks. These gaps are load-bearing for the claimed state-of-the-art margins, so the result cannot yet be accepted at face value.

major comments (4)
  1. [Section 3.1, Figures 4 and 11] The pseudo-GT concept maps G^k_x are used in the concept saliency alignment loss (Eq. 3), in the concept-pooled features f^k_x (Eq. 4), in the concept discovery criterion (Eq. 2), and in the concept signatures p_k (Eq. 6), yet the paper provides no quantitative evaluation of these maps. Figures 4 and 11 show only selected successful transfers, and Appendix A.4 adds more examples but no error rates. Because DIFT transfers annotations from a single synthesized exemplar per class to all real images across all domains, any systematic error in this transfer will corrupt the supervision signal and the stored signatures. Please add a quantitative validation of the annotation pipeline, for example IoU against human-annotated concept regions on a sample of images, or a DIFT correspondence benchmark, and/or an ablation in which the GT maps are corrupted or replaced to show the method's sensitivity to annotation quality.
  2. [Section 3.3, Algorithm 1, Eq. (4)] The test-time correction mechanism rests on an unverified premise. During training, concept features f^k_x are pooled using the ground-truth maps G^k_x (Eq. 4), and the signatures p_k are averages of these GT-pooled features (Eq. 6). At test time, Algorithm 1 (steps 4 and 14) replaces G^k_x with the model's predicted saliency maps S^k_x. The correction criterion d(f^k_x, p_k) > delta therefore only makes sense if S^k_x ≈ G^k_x on unseen target domains. The CSA loss (Eq. 3) enforces this alignment only on the source domain; no quantitative evidence shows it transfers. Since Table 2 attributes roughly 6.3 points of average PACS accuracy to test-time correction, this assumption is load-bearing. Please report quantitative alignment between predicted and pseudo-GT maps on target domains, and give per-domain correction success/failure statistics. The failure examples in Appendix A.1 are useful but do not replace rates.
  3. [Section 4, Table 1] For VLCS, OfficeHome, and DomainNet, the text states that the authors 'rely on reported results from respective papers, or compute them ourselves,' but it does not specify which entries were re-computed and which were copied from external papers. If the baseline numbers come from different backbones, train/validation splits, or evaluation protocols, the claimed margins of 13.37%, 16.16%, and 8.84% over the second-best approach may be inflated. Please provide per-method provenance, state explicitly which baselines were run under the same code base, and ideally re-run all baselines under a single protocol, or at least restrict the comparison claims to methods with identical experimental settings.
  4. [Section 3, Algorithm 1, Table 2] The method introduces several free parameters whose sensitivity is not analyzed: the concept importance threshold tau (Eq. 2), the LCC margin alpha (Eq. 5), the correction threshold delta and maximum iterations T (Algorithm 1), and the number/selection of concepts per class obtained from GPT-3.5. The main results use a single setting, so it is unclear how robust the reported gains and the correction behavior are to these choices. Please add ablations over tau, delta, alpha, and T, and report the concept sets used for each class in the supplementary material.
minor comments (5)
  1. [Section 3.2.2, Eq. (5)] The definition of the negative concept feature f^{k'}_{x^-} is implicit; please state explicitly how the negative image and its concept map are selected for each triplet, and whether Eq. (4) is applied using the negative's own GT map.
  2. [Algorithm 1] The control flow is ambiguous: the correction phase is nested inside the per-concept 'for' loop, and the return statements occur inside that loop. Please clarify whether each concept is checked independently and whether a corrected class must satisfy all concepts in K_c or only one.
  3. [Section 4, Experimental Setup] The description of the minimal augmentations ('quantization, blurring, and canny edge') and the positive/negative sampling rule for the triplets would benefit from exact parameter settings to make the LCC loss reproducible.
  4. [Figure 11 caption] The caption states that the figure shows concept saliency maps 'for ear and mouth,' but the displayed examples are a bird, a candle, a person, and a pig; please correct the caption to match the actual concepts shown.
  5. [Section 4, Table 1] The table reports point estimates without standard deviations or number of seeds. Since DomainBed typically reports mean accuracy over multiple runs, please state the number of seeds and, if possible, include standard deviations for the TIDE results.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: TIDE's training and test-time correction are not forced by construction, and the reported gains are evaluated against held-out target labels.

full rationale

The paper's derivation chain is self-contained in the sense required for a circularity finding. The concept-level ground-truth maps are produced by an external pipeline (GPT-3.5 concept selection, Stable Diffusion cross-attention maps, and DIFT transfer), not by the model being trained or evaluated. The training losses LCSA and LLCC are defined on source-domain data only, and the signatures p_k in Eq. 6 are averages of GT-pooled features from the training set. At test time, Algorithm 1 pools features using predicted saliency maps S^k_x and compares them with p_k; this is a heuristic self-consistency check whose validity depends on the empirical assumption that S^k_x approximates G^k_x after domain shift. That assumption is a correctness risk, not a circularity: no equation in the paper makes the comparison equal to its own input by construction, and the correction outcome is not a fitted parameter renamed as a prediction. Appendix A.1 explicitly documents cases where signature verification fails, which confirms the mechanism is not vacuous. Held-out target-domain labels provide external evaluation, and the qualitative saliency analyses are supplementary rather than load-bearing. The paper's self-citations (e.g., A-star for cross-attention, Sivaprasad et al. for DG framing) are contextual and do not carry the central argument. Accordingly, no circular step can be quoted with a specific reduction, and the circularity score is 0.

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

The method rests on the accuracy of automatically generated concept maps, the stability of concept signatures across domains, and the assumption that the source-domain classifier used for concept selection is a reliable guide. These are not validated quantitatively. The free parameters are largely set by hand, and the method introduces no new physical entities but relies on external models (GPT-3.5, SD v2.1, DIFT).

free parameters (5)
  • importance threshold tau = not specified
    Used in Eq. 2 to select important concepts per class; no value or selection procedure is reported.
  • LCC margin alpha = 1.0
    Set in Section 4 Experimental Setup; controls the triplet loss margin.
  • test-time correction threshold delta = 0.1
    Set empirically in Section 4; determines when concept-signature misalignment triggers correction.
  • max correction iterations T = 10
    Set empirically in Section 4; caps the iterative masking loop.
  • concept set per class = LLM-determined
    The set of concepts for each class is generated by GPT-3.5 and not fixed by any objective criterion.
assumptions (5)
  • domain assumption Diffusion model cross-attention maps accurately localize concepts in the synthesized exemplar images.
    Section 3.1 uses cross-attention maps as ground-truth concept localization for the generated exemplar, without quantitative validation.
  • domain assumption DIFT establishes reliable pixel-level correspondences between the synthetic exemplar and real images across all target domains.
    Section 3.1 transfers the exemplar saliency maps to real images via DIFT; no quantitative evaluation of transfer accuracy is provided.
  • ad hoc to paper The concept selection via GradCAM overlap (Eq. 1-2) identifies concepts that are truly discriminative and stable.
    The threshold tau is set implicitly, and the selection depends on a separately trained ResNet-18 classifier's attention, which may be biased.
  • domain assumption Concept signatures computed on the source domain remain valid for unseen target domains.
    Section 3.3.1 stores prototypes from source data; the test-time correction assumes these are stable across domains, which is exactly what domain generalization aims to achieve, making it a circular reliance on the method's own premise.
  • domain assumption Standard ResNet-18 architecture and training setup are sufficient for the claimed DG improvements.
    The method is demonstrated only with ResNet-18; no evidence for architecture generality.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TIDE: Training Locally Interpretable Domain Generalization Models Enables Test-time Correction." pith.science (2026). https://pith.science/paper/5SC7MVP3

@misc{pith2026241116788,
  author       = {Pith},
  title        = {Pith review of: TIDE: Training Locally Interpretable Domain Generalization Models Enables Test-time Correction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5SC7MVP3}},
  note         = {Machine review of arXiv:2411.16788}
}
read the original abstract

We consider the problem of single-source domain generalization. Existing methods typically rely on extensive augmentations to synthetically cover diverse domains during training. However, they struggle with semantic shifts (e.g., background and viewpoint changes), as they often learn global features instead of local concepts that tend to be domain invariant. To address this gap, we propose an approach that compels models to leverage such local concepts during prediction. Given no suitable dataset with per-class concepts and localization maps exists, we first develop a novel pipeline to generate annotations by exploiting the rich features of diffusion and large-language models. Our next innovation is TIDE, a novel training scheme with a concept saliency alignment loss that ensures model focus on the right per-concept regions and a local concept contrastive loss that promotes learning domain-invariant concept representations. This not only gives a robust model but also can be visually interpreted using the predicted concept saliency maps. Given these maps at test time, our final contribution is a new correction algorithm that uses the corresponding local concept representations to iteratively refine the prediction until it aligns with prototypical concept representations that we store at the end of model training. We evaluate our approach extensively on four standard DG benchmark datasets and substantially outperform the current state-ofthe-art (12% improvement on average) while also demonstrating that our predictions can be visually interpreted

Figures

Figures reproduced from arXiv: 2411.16788 by the authors.

Figure 1
Figure 1. Samples from VLCS (left) and PACS dataset (right) [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The TIDE pipeline: Left—Training on a single domain with cross-entropy losses for class ( [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The first column displays the image generated from the [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (7 more)
Figure 5
Figure 5. Figure 5: t-SNE visualizations to demonstrate impact of [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Illustrative examples of concept level GradCAM maps corresponding to TIDE’s predictions, across the four studied datasets. The [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: The top row shows initial class predictions and Grad [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Signature Verification Errors. Training Domain (Caltech101) Test Domain (LabelMe) Training Domain (Photo) Test Domain (Cartoon) NJPP Proposed bird person beak feathers eyes lips VLCS PACS [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 9
Figure 9. Figure 9: Samples from VLCS (left) and PACS dataset (right) [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: Additional Qualitative Results for Concept Localization. [PITH_FULL_IMAGE:figures/full_fig_p014_10.png]
Figure 11
Figure 11. Figure 11: For each dataset (PACS, VLCS, OfficeHome, and DomainNet), we show an example of concept saliency maps (for [PITH_FULL_IMAGE:figures/full_fig_p015_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

64 extracted references · 58 canonical work pages

  1. [1]

    A-star: Test-time attention segregation and retention for text-to-image synthesis

    Aishwarya Agarwal, Srikrishna Karanam, KJ Joseph, Apoorv Saxena, Koustava Goswami, and Balaji Vasan Srini- vasan. A-star: Test-time attention segregation and retention for text-to-image synthesis. In International Conference on Computer Vision (ICCV), 2023. 3, 6

    work page 2023

  2. [2]

    Language models are few-shot learners

    Tom B Brown. Language models are few-shot learners. Ad- vances in Neural Information Processing Systems (NeurIPS),

    work page

  3. [3]

    Attend-and-excite: Attention-based se- mantic guidance for text-to-image diffusion models

    Hila Chefer, Yuval Alaluf, Yael Vinker, Lior Wolf, and Daniel Cohen-Or. Attend-and-excite: Attention-based se- mantic guidance for text-to-image diffusion models. ACM Transactions on Graphics (TOG), 42(4):1–10, 2023. 3, 6

    work page 2023

  4. [4]

    Meta-causal learning for single domain generalization

    Jin Chen, Zhi Gao, Xinxiao Wu, and Jiebo Luo. Meta-causal learning for single domain generalization. In Conference on Computer Vision and Pattern Recognition (CVPR), 2023. 2, 6, 7

    work page 2023

  5. [5]

    Adversar- ial bayesian augmentation for single-source domain gener- alization

    Sheng Cheng, Tejas Gokhale, and Yezhou Yang. Adversar- ial bayesian augmentation for single-source domain gener- alization. In International Conference on Computer Vision (ICCV), 2023. 1, 2, 7

    work page 2023

  6. [6]

    Progressive random con- volutions for single domain generalization

    Seokeon Choi, Debasmit Das, Sungha Choi, Seunghan Yang, Hyunsin Park, and Sungrack Yun. Progressive random con- volutions for single domain generalization. InConference on Computer Vision and Pattern Recognition (CVPR), 2023. 6, 7

    work page 2023

  7. [7]

    Randaugment: Practical automated data augmentation with a reduced search space

    Ekin D Cubuk, Barret Zoph, Jonathon Shlens, and Quoc V Le. Randaugment: Practical automated data augmentation with a reduced search space. In Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2020. 1, 2, 6, 7

    work page 2020

  8. [8]

    Attention consistency on visual corruptions for single-source domain generalization

    Ilke Cugu, Massimiliano Mancini, Yanbei Chen, and Zeynep Akata. Attention consistency on visual corruptions for single-source domain generalization. InConference on Com- puter Vision and Pattern Recognition (CVPR), 2022. 1, 2, 6, 7

    work page 2022

Show all 64 references
  1. [9]

    Terrance Devries and Graham W. Taylor. Improved regular- ization of convolutional neural networks with cutout. ArXiv, abs/1708.04552, 2017. 7

    2017 arXiv

  2. [10]

    Learning to learn with variational information bottleneck for domain gener- alization

    Yingjun Du, Jun Xu, Huan Xiong, Qiang Qiu, Xiantong Zhen, Cees GM Snoek, and Ling Shao. Learning to learn with variational information bottleneck for domain gener- alization. In European Conference on Computer Vision (ECCV), 2020. 2

    2020

  3. [11]

    Domain gener- alization with domain-specific aggregation modules

    Antonio D’Innocente and Barbara Caputo. Domain gener- alization with domain-specific aggregation modules. In The German Conference on Pattern Recognition, 2019. 2

    2019

  4. [12]

    Adversarially adaptive normalization for single domain generalization

    Xinjie Fan, Qifei Wang, Junjie Ke, Feng Yang, Boqing Gong, and Mingyuan Zhou. Adversarially adaptive normalization for single domain generalization. In Conference on Com- puter Vision and Pattern Recognition (CVPR), 2021. 6, 7

    2021

  5. [13]

    Unbiased met- ric learning: On the utilization of multiple datasets and web images for softening bias

    Chen Fang, Ye Xu, and Daniel N Rockmore. Unbiased met- ric learning: On the utilization of multiple datasets and web images for softening bias. InConference on Computer Vision and Pattern Recognition (CVPR), 2013. 4, 6, 1

    2013

  6. [14]

    Imagenet-trained cnns are biased towards texture; increas- ing shape bias improves accuracy and robustness

    Robert Geirhos, Patricia Rubisch, Claudio Michaelis, Matthias Bethge, Felix A Wichmann, and Wieland Brendel. Imagenet-trained cnns are biased towards texture; increas- ing shape bias improves accuracy and robustness. In Inter- national Conference on Learning Representations (ICLR) ,

  7. [15]

    Domain generalization for object recog- nition with multi-task autoencoders

    Muhammad Ghifary, W Bastiaan Kleijn, Mengjie Zhang, and David Balduzzi. Domain generalization for object recog- nition with multi-task autoencoders. InInternational Confer- ence on Computer Vision (ICCV), 2015. 2

    2015

  8. [16]

    Attribute-guided adversarial training for robustness to nat- ural perturbations

    Tejas Gokhale, Rushil Anirudh, Bhavya Kailkhura, Ja- yaraman J Thiagarajan, Chitta Baral, and Yezhou Yang. Attribute-guided adversarial training for robustness to nat- ural perturbations. In Association for the Advancement of Artificial Intelligence (AAAI), 2021. 2

    2021

  9. [17]

    In search of lost do- main generalization

    Ishaan Gulrajani and David Lopez-Paz. In search of lost do- main generalization. International Conference on Learning Representations (ICLR), 2020. 6

    2020

  10. [18]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Conference on Computer Vision and Pattern Recognition (CVPR), 2016. 4

    2016

  11. [19]

    Augmix: A simple data processing method to improve robustness and uncertainty

    Dan Hendrycks, Norman Mu, Ekin D Cubuk, Barret Zoph, Justin Gilmer, and Balaji Lakshminarayanan. Augmix: A simple data processing method to improve robustness and uncertainty. International Conference on Learning Repre- sentations (ICLR), 2020. 1, 2, 6, 7

    2020

  12. [20]

    Pixmix: Dream- like pictures comprehensively improve safety measures

    Dan Hendrycks, Andy Zou, Mantas Mazeika, Leonard Tang, Bo Li, Dawn Song, and Jacob Steinhardt. Pixmix: Dream- like pictures comprehensively improve safety measures. In Conference on Computer Vision and Pattern Recognition (CVPR), 2022. 7

    2022

  13. [21]

    Self-challenging improves cross-domain generalization

    Zeyi Huang, Haohan Wang, Eric P Xing, and Dong Huang. Self-challenging improves cross-domain generalization. In European Conference on Computer Vision (ECCV), 2020. 1, 2, 6, 7

    2020

  14. [22]

    Spatial transformer networks

    Max Jaderberg, Karen Simonyan, Andrew Zisserman, et al. Spatial transformer networks. Advances in Neural Informa- tion Processing Systems (NeurIPS), 2015. 2

    2015

  15. [23]

    Concept bottleneck models

    Pang Wei Koh, Thao Nguyen, Yew Siang Tang, Stephen Mussmann, Emma Pierson, Been Kim, and Percy Liang. Concept bottleneck models. In International Conference on Machine Learning (ICML), 2020. 3

    2020

  16. [24]

    A review of domain adap- tation without target labels

    Wouter M Kouw and Marco Loog. A review of domain adap- tation without target labels. IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI) , 43(3):766–785,

  17. [25]

    Imagenet classification with deep convolutional neural net- works

    Alex Krizhevsky, Ilya Sutskever, and Geoffrey E Hinton. Imagenet classification with deep convolutional neural net- works. In Advances in Neural Information Processing Sys- tems (NeurIPS), 2012. 4

    2012

  18. [26]

    Deeper, broader and artier domain general- ization

    Da Li, Yongxin Yang, Yi-Zhe Song, and Timothy M Hospedales. Deeper, broader and artier domain general- ization. In International Conference on Computer Vision (ICCV), 2017. 4, 6, 1

    2017

  19. [27]

    Prompt-driven dynamic object-centric learning for single do- main generalization

    Deng Li, Aming Wu, Yaowei Wang, and Yahong Han. Prompt-driven dynamic object-centric learning for single do- main generalization. In Conference on Computer Vision and Pattern Recognition (CVPR), 2024. 2, 6, 7

    2024

  20. [28]

    Uncertainty modeling for out- of-distribution generalization

    Xiaotong Li, Yongxing Dai, Yixiao Ge, Jun Liu, Ying Shan, and Ling-Yu Duan. Uncertainty modeling for out- of-distribution generalization. International Conference on Learning Representations (ICLR), 2022. 2, 6, 7

    2022

  21. [29]

    Deep domain gener- alization via conditional invariant adversarial networks

    Ya Li, Xinmei Tian, Mingming Gong, Yajing Liu, Tongliang Liu, Kun Zhang, and Dacheng Tao. Deep domain gener- alization via conditional invariant adversarial networks. In European Conference on Computer Vision (ECCV), 2018. 1

    2018

  22. [30]

    Towards out-of-distribution generalization: A survey

    Jiashuo Liu, Zheyan Shen, Yue He, Xingxuan Zhang, Ren- zhe Xu, Han Yu, and Peng Cui. Towards out-of-distribution generalization: A survey. ArXiv, abs/2108.13624, 2021. 1

    2021 arXiv

  23. [31]

    Learning transferable features with deep adaptation net- works

    Mingsheng Long, Yue Cao, Jianmin Wang, and Michael Jor- dan. Learning transferable features with deep adaptation net- works. In International Conference on Machine Learning (ICML), 2015. 1

    2015

  24. [32]

    Do concept bot- tleneck models learn as intended? International Conference on Learning Representations (ICLR), 2021

    Andrei Margeloiu, Matthew Ashman, Umang Bhatt, Yanzhi Chen, Mateja Jamnik, and Adrian Weller. Do concept bot- tleneck models learn as intended? International Conference on Learning Representations (ICLR), 2021. 3

    2021

  25. [33]

    Unified deep supervised domain adap- tation and generalization

    Saeid Motiian, Marco Piccirilli, Donald A Adjeroh, and Gi- anfranco Doretto. Unified deep supervised domain adap- tation and generalization. In International Conference on Computer Vision (ICCV), 2017. 2

    2017

  26. [34]

    Domain generalization via invariant fea- ture representation

    Krikamol Muandet, David Balduzzi, and Bernhard Sch¨olkopf. Domain generalization via invariant fea- ture representation. In International Conference on Machine Learning (ICML), 2013. 2

    2013

  27. [35]

    Permuted adain: Reducing the bias towards global statistics in image clas- sification

    Oren Nuriel, Sagie Benaim, and Lior Wolf. Permuted adain: Reducing the bias towards global statistics in image clas- sification. In Conference on Computer Vision and Pattern Recognition (CVPR), 2021. 6, 7

    2021

  28. [36]

    Label-free concept bottleneck models

    Tuomas Oikarinen, Subhro Das, Lam M Nguyen, and Tsui- Wei Weng. Label-free concept bottleneck models. Inter- national Conference on Learning Representations (ICLR) ,

  29. [37]

    Moment matching for multi-source domain adaptation

    Xingchao Peng, Qinxun Bai, Xide Xia, Zijun Huang, Kate Saenko, and Bo Wang. Moment matching for multi-source domain adaptation. In International Conference on Com- puter Vision (ICCV), 2019. 4, 6, 1

    2019

  30. [38]

    Generalizing to unseen domains via text-guided augmenta- tion

    Daiqing Qi, Handong Zhao, Aidong Zhang, and Sheng Li. Generalizing to unseen domains via text-guided augmenta- tion. In European Conference on Computer Vision (ECCV),

  31. [39]

    Learning to learn single domain generalization

    Fengchun Qiao, Long Zhao, and Xi Peng. Learning to learn single domain generalization. In Conference on Computer Vision and Pattern Recognition (CVPR), 2020. 1

    2020

  32. [40]

    Modality-agnostic debiasing for single domain generalization

    Sanqing Qu, Yingwei Pan, Guang Chen, Ting Yao, Changjun Jiang, and Tao Mei. Modality-agnostic debiasing for single domain generalization. In Conference on Computer Vision and Pattern Recognition (CVPR), 2023. 2, 6, 7

    2023

  33. [41]

    High-resolution image syn- thesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj¨orn Ommer. High-resolution image syn- thesis with latent diffusion models. In Conference on Com- puter Vision and Pattern Recognition (CVPR), 2022. 2, 6

    2022

  34. [42]

    Grad-cam: Visual explanations from deep networks via gradient-based localization

    Ramprasaath R Selvaraju, Michael Cogswell, Abhishek Das, Ramakrishna Vedantam, Devi Parikh, and Dhruv Batra. Grad-cam: Visual explanations from deep networks via gradient-based localization. In International Conference on Computer Vision (ICCV), 2017. 1, 2, 3, 4

    2017

  35. [43]

    Class-wise domain generalization: A novel framework for evaluating distributional shift

    Sarath Sivaprasad, Akshay Goindani, Mario Fritz, and Vi- neet Gandhi. Class-wise domain generalization: A novel framework for evaluating distributional shift. In NeurIPS Workshop on Distribution Shifts: Connecting Methods and Applications, 2022. 1, 2

    2022

  36. [44]

    Frus- tratingly simple domain generalization via image stylization

    Nathan Somavarapu, Chih-Yao Ma, and Zsolt Kira. Frus- tratingly simple domain generalization via image stylization. ArXiv, abs/2006.11207, 2020. 2

    2006 arXiv

  37. [45]

    Class- imbalanced domain adaptation: An empirical odyssey

    Shuhan Tan, Xingchao Peng, and Kate Saenko. Class- imbalanced domain adaptation: An empirical odyssey. In European Conference on Computer Vision Workshops (EC- CVW), 2020. 6

    2020

  38. [46]

    Emergent correspondence from image diffusion

    Luming Tang, Menglin Jia, Qianqian Wang, Cheng Perng Phoo, and Bharath Hariharan. Emergent correspondence from image diffusion. In Advances in Neural Information Processing Systems (NeurIPS), 2023. 2, 4

    2023

  39. [47]

    Unbiased look at dataset bias

    Antonio Torralba and Alexei A Efros. Unbiased look at dataset bias. In Conference on Computer Vision and Pattern Recognition (CVPR), 2011. 2

    2011

  40. [48]

    Adversarial discriminative domain adaptation

    Eric Tzeng, Judy Hoffman, Kate Saenko, and Trevor Darrell. Adversarial discriminative domain adaptation. In Confer- ence on Computer Vision and Pattern Recognition (CVPR) ,

  41. [49]

    Deep hashing network for unsupervised domain adaptation

    Hemanth Venkateswara, Jose Eusebio, Shayok Chakraborty, and Sethuraman Panchanathan. Deep hashing network for unsupervised domain adaptation. In Conference on Com- puter Vision and Pattern Recognition (CVPR) , 2017. 4, 6, 1

    2017

  42. [50]

    Generalizing to unseen domains via adversarial data augmentation

    Riccardo V olpi, Hongseok Namkoong, Ozan Sener, John C Duchi, Vittorio Murino, and Silvio Savarese. Generalizing to unseen domains via adversarial data augmentation. Ad- vances in Neural Information Processing Systems (NeurIPS),

  43. [51]

    Mcpnet: An interpretable classifier via multi-level concept prototypes

    Bor-Shiun Wang, Chien-Yi Wang, and Wei-Chen Chiu. Mcpnet: An interpretable classifier via multi-level concept prototypes. In Conference on Computer Vision and Pattern Recognition (CVPR), 2024. 3

    2024

  44. [52]

    Learning robust representations by projecting super- ficial statistics out

    Haohan Wang, Zexue He, Zachary C Lipton, and Eric P Xing. Learning robust representations by projecting super- ficial statistics out. International Conference on Learning Representations (ICLR), 2019. 2

    2019

  45. [53]

    Generalizing to unseen domains: A survey on do- main generalization

    Jindong Wang, Cuiling Lan, Chang Liu, Yidong Ouyang, Tao Qin, Wang Lu, Yiqiang Chen, Wenjun Zeng, and S Yu Philip. Generalizing to unseen domains: A survey on do- main generalization. IEEE transactions on knowledge and data engineering, 35(8):8052–8072, 2022. 1

    2022

  46. [54]

    Learning to diversify for single do- main generalization

    Zijian Wang, Yadan Luo, Ruihong Qiu, Zi Huang, and Mahsa Baktashmotlagh. Learning to diversify for single do- main generalization. In International Conference on Com- puter Vision (ICCV), 2021. 6, 7

    2021

  47. [55]

    A bit more bayesian: Domain-invariant learning with uncertainty

    Zehao Xiao, Jiayi Shen, Xiantong Zhen, Ling Shao, and Cees Snoek. A bit more bayesian: Domain-invariant learning with uncertainty. In International Conference on Machine Learning (ICML), 2021. 2

    2021

  48. [56]

    Simde: A simple domain ex- pansion approach for single-source domain generalization

    Qinwei Xu, Ruipeng Zhang, Yi-Yan Wu, Ya Zhang, Ning Liu, and Yanfeng Wang. Simde: A simple domain ex- pansion approach for single-source domain generalization. In Conference on Computer Vision and Pattern Recognition (CVPR), 2023. 2

    2023

  49. [57]

    Language in a bottle: Language model guided concept bottlenecks for in- terpretable image classification

    Yue Yang, Artemis Panagopoulou, Shenghao Zhou, Daniel Jin, Chris Callison-Burch, and Mark Yatskar. Language in a bottle: Language model guided concept bottlenecks for in- terpretable image classification. In Conference on Computer Vision and Pattern Recognition (CVPR), 2023. 3

    2023

  50. [58]

    Not just pretty pictures: Toward interventional data augmentation using text-to-image generators

    Jianhao Yuan, Francesco Pinto, Adam Davies, and Philip Torr. Not just pretty pictures: Toward interventional data augmentation using text-to-image generators. International Conference on Machine Learning (ICML), 2024. 1, 2, 6, 7

    2024

  51. [59]

    Cutmix: Regu- larization strategy to train strong classifiers with localizable features

    Sangdoo Yun, Dongyoon Han, Seong Joon Oh, Sanghyuk Chun, Junsuk Choe, and Youngjoon Yoo. Cutmix: Regu- larization strategy to train strong classifiers with localizable features. In International Conference on Computer Vision (ICCV), 2019. 6, 7

    2019

  52. [60]

    Dauphin, and David Lopez-Paz

    Hongyi Zhang, Moustapha Cisse, Yann N. Dauphin, and David Lopez-Paz. mixup: Beyond empirical risk minimiza- tion. International Conference on Learning Representations (ICLR), 2018. 7

    2018

  53. [61]

    Towards principled disentanglement for domain generalization

    Hanlin Zhang, Yi-Fan Zhang, Weiyang Liu, Adrian Weller, Bernhard Sch ¨olkopf, and Eric P Xing. Towards principled disentanglement for domain generalization. In Conference on Computer Vision and Pattern Recognition (CVPR), 2022. 1

    2022

  54. [62]

    Maximum-entropy adversarial data augmentation for im- proved generalization and robustness

    Long Zhao, Ting Liu, Xi Peng, and Dimitris Metaxas. Maximum-entropy adversarial data augmentation for im- proved generalization and robustness. InAdvances in Neural Information Processing Systems (NeurIPS), 2020. 7

    2020

  55. [63]

    Do- main generalization with mixstyle

    Kaiyang Zhou, Yongxin Yang, Yu Qiao, and Tao Xiang. Do- main generalization with mixstyle. International Conference on Learning Representations (ICLR), 2021. 1, 2, 6, 7

    2021

  56. [64]

    Mixstyle neural networks for domain generalization and adaptation

    Kaiyang Zhou, Yongxin Yang, Yu Qiao, and Tao Xi- ang. Mixstyle neural networks for domain generalization and adaptation. International Journal of Computer Vision (IJCV), 132(3):822–836, 2024. 2 Appendix A. In Section A.1, we show examples where the proposed signature verificat...

    2024

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.