Pith. sign in

REVIEW 3 major objections 6 minor 72 references

Text-Driven Causal Representation Learning for Source-Free Domain Generalization

T0 review · 3 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A causal do-operation on the style part of CLIP text embeddings removes domain confounders, letting a network trained with InfoNCE beat prior source-free domain generalization on four benchmarks.

desk verdict Useful SFDG method with consistent gains, but the causal theory does not match the implementation: the InfoNCE-trained network is style-specific, so the headline do-operation claim is overstated. read the letter →

arxiv 2507.09961 v1 pith:ZSEYJUEV submitted 2025-07-14 cs.LG

Lihua Zhou , Mao Ye , Nianxin Li , Shuaifeng Li , Jinlin Wu , Xiatian Zhu , Lei Deng , Hongbin Liu
show 2 more authors
Jiebo Luo Zhen Lei
This is my paper · ORCID
classification cs.LG
keywords source-freedomaingeneralizationcausalrepresentationlearningdo-operationCLIPtextembeddingsconfounderdictionarydomain-invariantfeaturescontrastivevision-languagemodels
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

Source-free domain generalization aims to classify images from unseen domains using only class names and a pre-trained vision-language model, without any source images. This paper proposes TDCRL, which synthesizes training samples as text embeddings from prompts like 'a cat in a sketch style' and treats the style part as a domain-specific confounder. The paper claims and proves, under an independence assumption, that a causal do-operation on the style component recovers the class-only distribution, i.e., $P(Y|\mathrm{do}(f_{cls}^{s_m})) = P(Y|f_{cls})$. It then trains an intervention network with a contrastive loss to realize this operation and reports state-of-the-art average accuracy on PACS, VLCS, OfficeHome, and DomainNet without source-domain data.

What carries the argument

The load-bearing objects are the do-operation from causal inference, the confounder dictionary $Z = \{z_n\}$, and the intervention network $g(\cdot)$. The do-operation formalizes replacing the style component $f_{s_m}$ of a text embedding with a chosen style vector $z_n$, expressing the intervened label distribution as an expectation over $Z$. The confounder dictionary is a set of $N$ style embeddings produced from prompts like 'a object in a sketch style'; importantly, $N$ is finite in practice even though the derivation treats it as infinite. The intervention network $g(\cdot)$ is a small fully connected network that takes a text embedding and a style vector as inputs and outputs an intervened feature; it is trained so that $g(f_{cls}^{s_m}, z_n)$ matches the embedding of 'a [class] in a [domain] style' for the same class and style while being pushed away from other styles by the InfoNCE loss. Finally, the Normalized Weighted Geometric Mean approximation moves the expectation inside the softmax, so the classifier consumes the average intervened feature $\frac{1}{N}\sum_{n=1}^{N} g(f_{cls}^{s_m}, z_n)$. This combination is what carries the argument from the causal identity to a trainable classifier.

What would settle it

Compute the average cosine similarity (or mutual information) between the class-feature embeddings and the style-intervention vectors for all class-style pairs in the paper's prompt template; if the similarity is clearly above the level seen between unrelated CLIP embeddings, the independence assumption in Eq. (4) is violated. More directly, train the classifier on intervened features and compare its output distribution against CLIP's prediction from class-only prompts on the same test images; a systematic divergence between the two would show that the claimed identity is not what the network realizes.

Watch

Extended reading notes

Core claim

The central claim is that a causal intervention on the style component of a CLIP text embedding removes domain-specific confounding and recovers the distribution of labels given the class feature alone. Concretely, Eq. (7) states $P(Y|\mathrm{do}(f_{cls}^{s_m})) = P(Y|f_{cls})$, where $f_{cls}^{s_m}$ is a text embedding combining a class feature $f_{cls}$ with a style feature $f_{s_m}$. The derivation introduces a confounder dictionary $Z = \{z_n\}$ of $N$ style intervention vectors, replaces the style component with each $z_n$, and sums over the dictionary; the key step is the independence assumption $P(f_{cls}, z_n) = P(f_{cls})P(z_n)$. The paper implements this by training a neural network $g(\cdot)$ with an InfoNCE contrastive loss to map $f_{cls}^{s_m}$ to the intervened feature $g(f_{cls}^{s_m}, z_n)$, and then training a classifier on the expectation over all $z_n$. The same $g(\cdot)$ is applied to image features at test time, transferring the text-learned intervention across modalities.

Load-bearing premise

The derivation rests on the assumption that class identity and style are statistically independent in the text embedding, so that the joint probability factors as $P(f_{cls}, z_n) = P(f_{cls})P(z_n)$; if real embeddings correlate the two, the equality $P(Y|\mathrm{do}(f_{cls}^{s_m})) = P(Y|f_{cls})$ does not follow.

Editorial extensions

If this is right

  • If the identity holds, classifiers trained on intervened text embeddings are effectively trained on pure class features, so style cannot act as a confounder on unseen test domains.
  • The same intervention network transfers from text-trained features to image features at test time, which is what makes source-free generalization possible without any adaptation on the target domain.
  • The reported gains are largest on smaller backbones, which suggests the causal intervention substitutes for the robustness that larger pre-trained models already provide.
  • Style information is measurably removed after intervention: style-classification cross-entropy rises and average MMD distance between domain pairs falls on all four benchmarks.
  • Ablations show the intervention contributes roughly 0.7 to 3.4 percentage points of accuracy across backbones, and accuracy improves with the number of confounders up to a point of diminishing returns.

Reading between the lines

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

  • The independence assumption $P(f_{cls}, z_n) = P(f_{cls})P(z_n)$ is plausible for synthetic prompts, where class and style tokens are concatenated independently, but it is unlikely to hold exactly for real image features; a variant that models the dependence between class and style is a direct extension.
  • The implemented model uses the softmax of an expected feature rather than the expected softmax, so it is worth measuring how much of the causal guarantee survives the Normalized Weighted Geometric Mean approximation on these benchmarks.
  • Because the confounder dictionary is built from a fixed list of style words, the framework naturally extends to unseen styles by appending new style words and recomputing embeddings, without retraining the backbone.
  • A direct test of the central identity would be to compare classifier predictions on intervened features against predictions made from class-only prompts on the same test images; a systematic divergence would indicate that the style intervention is not fully removing the confounder.
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 / 6 minor

Summary. TDCRL is a source-free domain generalization method built on CLIP text embeddings. The method first generates diverse style word vectors by mixup and random sampling from D predefined domain words, constructs M×K text prompts of the form "a [class_k] in a [style_m] style", and encodes them with CLIP's text encoder to simulate visual embeddings. It then trains a causal intervention network g(·) together with a confounder dictionary Z={z_n}_{n=1}^N using an InfoNCE loss, and trains a linear classifier on the average intervened feature (1/N)∑_n g(f,z_n). The paper gives a theoretical derivation (Eqs. 2–7) claiming that P(Y|do(f_cls_sm))=P(Y|f_cls), and reports state-of-the-art accuracy on PACS, VLCS, OfficeHome, and DomainNet with ResNet-50, ViT-B/16, and ViT-L/14 backbones, together with ablations on loss functions, confounder dictionary size, number of layers, and hyperparameters.

Significance. If the causal interpretation were sound, TDCRL would be a practical and efficient causal SFDG method with consistent gains over a broad set of baselines, including source-dependent methods. The empirical evaluation is substantial: three backbones, four benchmarks, comparisons with recent CLIP-based methods, component ablations, loss-function ablations, sensitivity analyses, t-SNE visualization, and cross-modal invariance metrics. However, the central theoretical claim is not established. The derivation assumes independence between class features and style interventions (Eq. 4) and treats a six-entry dictionary as an exhaustive style partition (Eq. 6), while the training objective in Eq. (11) explicitly rewards style-specific outputs. The empirical contribution is still potentially valuable as a text-driven feature-augmentation and regularization method, but the causal guarantee is currently unsupported.

major comments (3)
  1. [III-B1, Eqs. (4), (6), and (7)] The proof of the central claim P(Y|do(f_cls_sm))=P(Y|f_cls) rests on two unjustified steps. First, Eq. (4) asserts f_cls⊥z_n, the independence of the category feature and the style intervention vector; no argument or measurement is given for why this holds for real or even synthesized image features. Second, Eq. (6) requires that summing the joint P(Y,f_cls,z_n) over the N entries of Z yields the marginal P(Y,f_cls), which is only valid if the dictionary is an exhaustive partition of the style space. The paper selects N=6 words out of D=13 predefined domain words (Section IV), which is an arbitrary subset, not a partition of all possible styles. Consequently Eq. (7) does not follow from the preceding equations.
  2. [III-B2, Eqs. (10)-(12)] Even if Eqs. (2)-(7) were accepted, the trained network does not implement the object described by the theorem. Eq. (11) trains g(f_cls_sm,z_n) to match the embedding F[k,n,:] of "a [class_k] in a [domain_n] style" and to be dissimilar from F[k,j,:] for j≠n, so each intervened feature is explicitly style-specific. The classifier is then evaluated on (1/N)∑_n g(f_cls_sm,z_n), and nothing in the objective enforces that this average equals a class-only feature f_cls or that the label distribution is invariant to z_n, i.e., P(Y|f_cls,z_n)=P(Y|f_cls). The NWGM step in Eq. (10) only moves an expectation through a softmax and cannot repair this mismatch. The paper should either add an explicit invariance constraint that realizes the claimed intervention or reframe the method as a causal-inspired heuristic rather than a method with a proven equivalence.
  3. [IV, Tables VIII and Fig. 4] Key hyperparameters, including the confounder dictionary size N=6, the number of layers in g(·), the temperature τ=0.1, and the balance weight λ=3.0, appear to be selected using accuracy on the same test benchmarks reported in Table I. Tables VII and VIII and Fig. 4 show the effect of these choices on PACS, VLCS, OfficeHome, and DomainNet. If the final configuration was chosen after inspecting performance on the test domains, the reported state-of-the-art numbers are optimistic and the margins over the best baselines are small (e.g., 0.1–0.6 points in many cells). A clearly described validation protocol, such as tuning on a held-out split of the source text embeddings or a separate validation set, is needed to support the empirical claim.
minor comments (6)
  1. [III-A, Eq. (1)] The Gaussian noise term ϵ in Eq. (1) is introduced in word-vector space, but its magnitude and scaling are not specified; please clarify whether it is normalized and whether it can violate the simplex constraint on the mixup weights.
  2. [III-B2, Eq. (8)] The notation "f_sm → z_n" (and "f_sm −→ z_n" in Eq. (8)) is undefined; use a standard do-notation or explicitly define the intervention operator before first use.
  3. [IV, Implementation details] There are several typos: "N = 6 style world" should be "N = 6 style words", "a object in a ... style" should be "an object ...", and "M = 80style word vectors" is missing a space. Figure 3 also says "Confunder Z" instead of "Confounder Z."
  4. [III-B2, Eq. (10)] The Normalized Weighted Geometric Mean approximation is cited to [59], which is an image-captioning paper; please cite the original source of the NWGM approximation or a standard reference for it.
  5. [IV-A, Table I] The text states that TDCRL achieves state-of-the-art results, but on DomainNet with ResNet-50 TDCRL obtains 49.8%, which is 0.3 points below SPG's 50.1%; the claim should be qualified to acknowledge this exception.
  6. [IV-B, Feature visualization] The t-SNE plot in Fig. 5 is descriptive and qualitative; consider supplementing it with a quantitative invariance metric computed on held-out target domains, such as domain classification accuracy or MMD with a separate validation split.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the causal equivalence is derived from explicitly stated independence and exhaustiveness assumptions, and the performance claims are evaluated against external benchmarks.

full rationale

The paper's central equivalence P(Y|do(f_cls_sm)) = P(Y|f_cls) (Eq. 7) is an algebraic consequence of two explicit assumptions: f_cls is independent of z_n (Eq. 4) and the style dictionary is exhaustive so the total-probability summation holds (Eq. 6). These assumptions are stated before the derivation and are not obtained from the conclusion, so the derivation is not circular in the definitional sense; whether the independence holds for real CLIP features is a validity/correctness risk, not a circularity. The implementation trains g(·) to map (f_cls_sm, z_n) to F[k,n,:] and averages over the N dictionary entries at prediction time, which is consistent with the expectation in Eq. (10); whether this truly removes confounders is an empirical question, and the paper evaluates it on external benchmarks (PACS, VLCS, OfficeHome, DomainNet) plus style-classification loss and MMD analyses. The acknowledged limitation that N is finite in practice ('we can only regard it as a finite number') weakens Eq. (6) but is an approximation rather than a fitted parameter renamed as a prediction. No load-bearing self-citations or imported uniqueness theorems appear; the citations to Pearl, IRM, and DPStyler are standard external references. The main concerns are therefore the unverified independence assumption and the finite-dictionary approximation, not circular reasoning.

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

The method relies on several hand-chosen hyperparameters and a load-bearing independence assumption. The confounder dictionary and intervention network are method components rather than newly postulated physical entities.

free parameters (5)
  • Domain word list D=13 = sketch, cartoon, photo, surrealism, minimalist, retro, pixel-art, collage, pointillism, stained-glass, illustration…
    Hand-picked words intended to span benchmark visual styles; used both to generate M=80 augmented style vectors and to construct the confounder dictionary.
  • Confounder dictionary size N = 6
    Selected via ablation on the four test benchmarks (Table VIII); larger N gives diminishing returns but increases compute.
  • Temperature tau = 0.1
    Selected via sensitivity analysis on PACS and DomainNet test sets (Fig. 4).
  • Balance lambda = 3.0
    Selected via sensitivity analysis on PACS and DomainNet test sets (Fig. 4).
  • Number of layers in g = 3
    Selected via ablation on test benchmarks (Table VII); more layers give marginal gains but slower inference.
assumptions (4)
  • domain assumption CLIP text and image embeddings share a common space, so text can simulate visual representations.
    State-of-the-art SFDG relies on this; invoked throughout Section III-A.
  • ad hoc to paper f_cls and z_n are independent (Eq. 4).
    Needed for the equivalence proof; not empirically validated for real images.
  • domain assumption The do-operation formula (Eq. 2) correctly models causal intervention in this setting.
    The backdoor-style sum is written down without a formal structural causal model; its validity for this setting is assumed.
  • standard math NWGM approximation is acceptable for softmax expectation.
    Used in Eq. (10) to swap expectation and softmax; standard but only approximate.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Text-Driven Causal Representation Learning for Source-Free Domain Generalization." pith.science (2026). https://pith.science/paper/ZSEYJUEV

@misc{pith2026250709961,
  author       = {Pith},
  title        = {Pith review of: Text-Driven Causal Representation Learning for Source-Free Domain Generalization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZSEYJUEV}},
  note         = {Machine review of arXiv:2507.09961}
}
read the original abstract

Deep learning often struggles when training and test data distributions differ. Traditional domain generalization (DG) tackles this by including data from multiple source domains, which is impractical due to expensive data collection and annotation. Recent vision-language models like CLIP enable source-free domain generalization (SFDG) by using text prompts to simulate visual representations, reducing data demands. However, existing SFDG methods struggle with domain-specific confounders, limiting their generalization capabilities. To address this issue, we propose TDCRL (\textbf{T}ext-\textbf{D}riven \textbf{C}ausal \textbf{R}epresentation \textbf{L}earning), the first method to integrate causal inference into the SFDG setting. TDCRL operates in two steps: first, it employs data augmentation to generate style word vectors, combining them with class information to generate text embeddings to simulate visual representations; second, it trains a causal intervention network with a confounder dictionary to extract domain-invariant features. Grounded in causal learning, our approach offers a clear and effective mechanism to achieve robust, domain-invariant features, ensuring robust generalization. Extensive experiments on PACS, VLCS, OfficeHome, and DomainNet show state-of-the-art performance, proving TDCRL effectiveness in SFDG.

Figures

Figures reproduced from arXiv: 2507.09961 by the authors.

Figure 1
Figure 1. Comparison of DG and SFDG. (a) DG relies on expensive image [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. When training a classifier using observed [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Overview of our proposed TDCRL. (a) Text Embedding Generation: Starting with D predefined domain words, we use mixup and random sampling to generate M style word vectors {sm}M m=1 (M > D), creating M ×K text embeddings {{f clsk sm }K k=1}M m=1 via CLIP’s text encoder T(·) as training samples. (b) Causal Representation Learning: A confounder dictionary Z of N style embeddings and ground truth F, both from T(·), guide… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Sensitivity analysis with respect to τ /λ on PACS and DomainNet using ViT-B/16 as the visual backbone. robust to loss function choice. This stability suggests that the causal intervention mechanism, central to TDCRL, effectively extracts domain-invariant features regar…
Figure 5
Figure 5. Figure 5: t-SNE visualization of ”dog” category features in PACS using ViT [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

72 extracted references · 62 canonical work pages

  1. [1]

    Domain general- ization: A survey,

    K. Zhou, Z. Liu, Y . Qiao, T. Xiang, and C. C. Loy, “Domain general- ization: A survey,” IEEE transactions on pattern analysis and machine intelligence, vol. 45, no. 4, pp. 4396–4415, 2022

    work page 2022

  2. [2]

    Generalizing to unseen domains: A survey on domain generalization,

    J. Wang, C. Lan, C. Liu, Y . Ouyang, T. Qin, W. Lu, Y . Chen, W. Zeng, and S. Y . Philip, “Generalizing to unseen domains: A survey on domain generalization,” IEEE transactions on knowledge and data engineering , vol. 35, no. 8, pp. 8052–8072, 2022

    work page 2022

  3. [3]

    Multi-scale part-based feature representation for 3d domain generalization and adaptation,

    X. Wei, X. Gu, and J. Sun, “Multi-scale part-based feature representation for 3d domain generalization and adaptation,” IEEE Transactions on Pattern Analysis and Machine Intelligence , 2024

    work page 2024

  4. [4]

    Rotation-adaptive point cloud domain generalization via intricate orientation learning,

    B. Liu, C. Zheng, X. Xu, C. Xu, H. Zhang, and S. He, “Rotation-adaptive point cloud domain generalization via intricate orientation learning,” IEEE Transactions on Pattern Analysis and Machine Intelligence , 2025

    work page 2025

  5. [5]

    Incorporating pre-training data matters in unsupervised domain adaptation,

    Y . Xu, A. Men, Y . Liu, X. Zhuang, and Q. Chen, “Incorporating pre-training data matters in unsupervised domain adaptation,” IEEE Transactions on Pattern Analysis and Machine Intelligence , 2025

    work page 2025

  6. [6]

    Unified domain adaptive semantic segmentation,

    Z. Zhang, G. Wu, J. Zhang, X. Zhu, D. Tao, and T. Chai, “Unified domain adaptive semantic segmentation,” IEEE transactions on pattern analysis and machine intelligence , 2025

    work page 2025

  7. [7]

    Unraveling the mysteries of label noise in source-free domain adaptation: Theory and practice,

    G. Xu, L. Yi, P. Xu, J. Li, R. Pu, C. Shui, C. Ling, A. I. McLeod, and B. Wang, “Unraveling the mysteries of label noise in source-free domain adaptation: Theory and practice,” IEEE Transactions on Pattern Analysis and Machine Intelligence , 2025

    work page 2025

  8. [8]

    Invariant risk minimization,

    M. Arjovsky, L. Bottou, I. Gulrajani, and D. Lopez-Paz, “Invariant risk minimization,” arXiv preprint arXiv:1907.02893 , 2019

    arXiv 1907

Show all 72 references
  1. [9]

    Domain generalization with mixstyle,

    K. Zhou, Y . Yang, Y . Qiao, and T. Xiang, “Domain generalization with mixstyle,” in International Conference on Learning Representations

  2. [10]

    Learning transferable visual models from natural language supervision,

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

    2021

  3. [11]

    Scaling up visual and vision-language representation learning with noisy text supervision,

    C. Jia, Y . Yang, Y . Xia, Y .-T. Chen, Z. Parekh, H. Pham, Q. Le, Y .-H. Sung, Z. Li, and T. Duerig, “Scaling up visual and vision-language representation learning with noisy text supervision,” in International conference on machine learning . PMLR, 2021, pp. 4904–4916

    2021

  4. [12]

    Vision-language pre-training with triple contrastive learning,

    J. Yang, J. Duan, S. Tran, Y . Xu, S. Chanda, L. Chen, B. Zeng, T. Chilimbi, and J. Huang, “Vision-language pre-training with triple contrastive learning,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 15 671–15 680

    2022

  5. [13]

    Modeling caption diversity in contrastive vision-language pretraining,

    S. Lavoie, P. Kirichenko, M. Ibrahim, M. Assran, A. G. Wilson, A. Courville, and N. Ballas, “Modeling caption diversity in contrastive vision-language pretraining,” in Forty-first International Conference on Machine Learning

  6. [14]

    Vision-language models for vision tasks: A survey,

    J. Zhang, J. Huang, S. Jin, and S. Lu, “Vision-language models for vision tasks: A survey,” IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024

    2024

  7. [15]

    Promptstyler: Prompt-driven style generation for source-free domain generalization,

    J. Cho, G. Nam, S. Kim, H. Yang, and S. Kwak, “Promptstyler: Prompt-driven style generation for source-free domain generalization,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 15 702–15 712

    2023

  8. [16]

    Promptta: Prompt- driven text adapter for source-free domain generalization,

    H. Zhang, S. Bai, W. Zhou, J. Fu, and B. Chen, “Promptta: Prompt- driven text adapter for source-free domain generalization,”arXiv preprint arXiv:2409.14163, 2024

    2024 arXiv

  9. [17]

    Dpstyler: dynamic promptstyler for source-free domain generalization,

    Y . Tang, Y . Wan, L. Qi, and X. Geng, “Dpstyler: dynamic promptstyler for source-free domain generalization,” IEEE Transactions on Multime- dia, 2025

    2025

  10. [18]

    Batstyler: Advancing multi- category style generation for source-free domain generalization,

    X. Xu, L. Qi, J. Zhou, and X. Geng, “Batstyler: Advancing multi- category style generation for source-free domain generalization,” IEEE Transactions on Circuits and Systems for Video Technology , 2025

    2025

  11. [19]

    Models, reasoning and inference,

    J. Pearl et al. , “Models, reasoning and inference,” Cambridge, UK: CambridgeUniversityPress, vol. 19, no. 2, p. 3, 2000

    2000

  12. [20]

    Causality for machine learning,

    B. Sch ¨olkopf, “Causality for machine learning,” in Probabilistic and causal inference: The works of Judea Pearl , 2022, pp. 765–804

    2022

  13. [21]

    Causality inspired representation learning for domain generalization,

    F. Lv, J. Liang, S. Li, B. Zang, C. H. Liu, Z. Wang, and D. Liu, “Causality inspired representation learning for domain generalization,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 8046–8056

    2022

  14. [22]

    Causality-inspired single-source domain generalization for medical image segmentation,

    C. Ouyang, C. Chen, S. Li, Z. Li, C. Qin, W. Bai, and D. Rueckert, “Causality-inspired single-source domain generalization for medical image segmentation,” IEEE Transactions on Medical Imaging , vol. 42, no. 4, pp. 1095–1106, 2022

    2022

  15. [23]

    Domain generalization via invariant feature representation,

    K. Muandet, D. Balduzzi, and B. Sch ¨olkopf, “Domain generalization via invariant feature representation,” in International conference on machine learning. PMLR, 2013, pp. 10–18

    2013

  16. [24]

    Robust domain generalisation by enforcing distribution invariance,

    S. Erfani, M. Baktashmotlagh, M. Moshtaghi, X. Nguyen, C. Leckie, J. Bailey, and R. Kotagiri, “Robust domain generalisation by enforcing distribution invariance,” in Proceedings of the Twenty-Fifth International Joint Conference on Artificial Intelligence (IJCAI-16) . AAAI Pre...

    2016

  17. [25]

    Unified deep supervised domain adaptation and generalization,

    S. Motiian, M. Piccirilli, D. A. Adjeroh, and G. Doretto, “Unified deep supervised domain adaptation and generalization,” in Proceedings of the IEEE international conference on computer vision , 2017, pp. 5715– 5725

    2017

  18. [26]

    Generalizable feature learning in the presence of data bias and domain class imbalance with application to skin lesion classification,

    C. Yoon, G. Hamarneh, and R. Garbi, “Generalizable feature learning in the presence of data bias and domain class imbalance with application to skin lesion classification,” in Medical Image Computing and Computer Assisted Intervention–MICCAI 2019: 22nd International Conference...

    2019

  19. [27]

    Respecting domain relations: Hypothesis invariance for domain generalization,

    Z. Wang, M. Loog, and J. Van Gemert, “Respecting domain relations: Hypothesis invariance for domain generalization,” in 2020 25th Inter- national Conference on Pattern Recognition (ICPR) . IEEE, 2021, pp. 9756–9763

    2020

  20. [28]

    Domain generalization for medical imaging classification with linear-dependency regularization,

    H. Li, Y . Wang, R. Wan, S. Wang, T.-Q. Li, and A. C. Kot, “Domain generalization for medical imaging classification with linear-dependency regularization,” in 34th Conference on Neural Information Processing Systems (NeurIPS 2020) , 2020

    2020

  21. [29]

    Domain generalization with adversarial feature learning,

    H. Li, S. J. Pan, S. Wang, and A. C. Kot, “Domain generalization with adversarial feature learning,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2018, pp. 5400–5409

    2018

  22. [30]

    Deep domain generalization via conditional invariant adversarial networks,

    Y . Li, X. Tian, M. Gong, Y . Liu, T. Liu, K. Zhang, and D. Tao, “Deep domain generalization via conditional invariant adversarial networks,” in Proceedings of the European conference on computer vision (ECCV) , 2018, pp. 624–639

    2018

  23. [31]

    Multi-adversarial discriminative deep domain generalization for face presentation attack detection,

    R. Shao, X. Lan, J. Li, and P. C. Yuen, “Multi-adversarial discriminative deep domain generalization for face presentation attack detection,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 10 023–10 031

    2019

  24. [32]

    Correlation-aware adversarial domain adaptation and generalization,

    M. M. Rahman, C. Fookes, M. Baktashmotlagh, and S. Sridharan, “Correlation-aware adversarial domain adaptation and generalization,” Pattern Recognition, vol. 100, p. 107124, 2020. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 12

    2020

  25. [33]

    Learning to generalize: Meta-learning for domain generalization,

    D. Li, Y . Yang, Y .-Z. Song, and T. Hospedales, “Learning to generalize: Meta-learning for domain generalization,” in Proceedings of the AAAI conference on artificial intelligence , vol. 32, no. 1, 2018

    2018

  26. [34]

    Metareg: towards domain generalization using meta-regularization,

    Y . Balaji, S. Sankaranarayanan, and R. Chellappa, “Metareg: towards domain generalization using meta-regularization,” in Proceedings of the 32nd International Conference on Neural Information Processing Systems, 2018, pp. 1006–1016

    2018

  27. [35]

    Metanorm: Learning to nor- malize few-shot batches across domains,

    Y . Du, X. Zhen, L. Shao, and C. G. Snoek, “Metanorm: Learning to nor- malize few-shot batches across domains,” in International Conference on Learning Representations , 2020

    2020

  28. [36]

    Towards univer- sal representation learning for deep face recognition,

    Y . Shi, X. Yu, K. Sohn, M. Chandraker, and A. K. Jain, “Towards univer- sal representation learning for deep face recognition,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2020, pp. 6817–6826

    2020

  29. [37]

    Addressing model vulnerability to distribu- tional shifts over image transformation sets,

    R. V olpi and V . Murino, “Addressing model vulnerability to distribu- tional shifts over image transformation sets,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2019, pp. 7980–7989

    2019

  30. [38]

    Generalizing to unseen domains via adversarial data augmentation,

    R. V olpi, H. Namkoong, O. Sener, J. Duchi, V . Murino, and S. Savarese, “Generalizing to unseen domains via adversarial data augmentation,” in Proceedings of the 32nd International Conference on Neural Information Processing Systems, 2018, pp. 5339–5349

    2018

  31. [39]

    Learning to learn single domain gen- eralization,

    F. Qiao, L. Zhao, and X. Peng, “Learning to learn single domain gen- eralization,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2020, pp. 12 556–12 565

    2020

  32. [40]

    Domain generalization by solving jigsaw puzzles,

    F. M. Carlucci, A. D’Innocente, S. Bucci, B. Caputo, and T. Tommasi, “Domain generalization by solving jigsaw puzzles,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2019, pp. 2229–2238

    2019

  33. [41]

    Self-supervised learning across domains,

    S. Bucci, A. D’Innocente, Y . Liao, F. M. Carlucci, B. Caputo, and T. Tommasi, “Self-supervised learning across domains,” IEEE Trans- actions on Pattern Analysis and Machine Intelligence , vol. 44, no. 9, pp. 5516–5528, 2021

    2021

  34. [42]

    Learning from extrin- sic and intrinsic supervisions for domain generalization,

    S. Wang, L. Yu, C. Li, C.-W. Fu, and P.-A. Heng, “Learning from extrin- sic and intrinsic supervisions for domain generalization,” in European Conference on Computer Vision . Springer, 2020, pp. 159–176

    2020

  35. [43]

    Filip: Fine-grained interactive language-image pre-training,

    L. Yao, R. Huang, L. Hou, G. Lu, M. Niu, H. Xu, X. Liang, Z. Li, X. Jiang, and C. Xu, “Filip: Fine-grained interactive language-image pre-training,” in International Conference on Learning Representations

  36. [44]

    Pali: A jointly-scaled multilingual language-image model,

    X. Chen, X. Wang, S. Changpinyo, A. Piergiovanni, P. Padlewski, D. Salz, S. Goodman, A. Grycner, B. Mustafa, L. Beyer et al. , “Pali: A jointly-scaled multilingual language-image model,” in The Eleventh International Conference on Learning Representations

  37. [45]

    Integrating neural-symbolic reasoning with variational causal inference network for explanatory visual ques- tion answering,

    D. Xue, S. Qian, and C. Xu, “Integrating neural-symbolic reasoning with variational causal inference network for explanatory visual ques- tion answering,” IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024

    2024

  38. [46]

    A causal adjustment module for debiasing scene graph generation,

    L. Liu, S. Sun, S. Zhi, F. Shi, Z. Liu, J. Heikkil ¨a, and Y . Liu, “A causal adjustment module for debiasing scene graph generation,” IEEE Transactions on Pattern Analysis and Machine Intelligence , 2025

    2025

  39. [47]

    Intersecting the markov blankets of endogenous and exogenous variables for causal discovery,

    Y . Dong and C. Gao, “Intersecting the markov blankets of endogenous and exogenous variables for causal discovery,” IEEE Transactions on Pattern Analysis and Machine Intelligence , 2025

    2025

  40. [48]

    A causal perspective on dataset bias in machine learning for medical imaging,

    C. Jones, D. C. Castro, F. De Sousa Ribeiro, O. Oktay, M. McCradden, and B. Glocker, “A causal perspective on dataset bias in machine learning for medical imaging,” Nature Machine Intelligence , vol. 6, no. 2, pp. 138–146, 2024

    2024

  41. [49]

    Visual commonsense r- cnn,

    T. Wang, J. Huang, H. Zhang, and Q. Sun, “Visual commonsense r- cnn,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2020, pp. 10 760–10 770

    2020

  42. [50]

    Causal imitation learning with unobserved confounders,

    J. Zhang, D. Kumor, and E. Bareinboim, “Causal imitation learning with unobserved confounders,” in Advances in neural information processing systems, 2020, pp. 12 263–12 274

    2020

  43. [51]

    Woulda, coulda, shoulda: Counterfactually-guided policy search,

    L. Buesing, T. Weber, Y . Zwols, N. Heess, S. Racaniere, A. Guez, and J.-B. Lespiau, “Woulda, coulda, shoulda: Counterfactually-guided policy search,” in International Conference on Learning Representations, 2018

    2018

  44. [52]

    Causal inference in natural language processing: Estimation, prediction, interpretation and beyond,

    A. Feder, K. A. Keith, E. Manzoor, R. Pryzant, D. Sridhar, Z. Wood- Doughty, J. Eisenstein, J. Grimmer, R. Reichart, M. E. Roberts et al. , “Causal inference in natural language processing: Estimation, prediction, interpretation and beyond,” Transactions of the Association for...

    2022

  45. [53]

    Domain generalization using causal matching,

    D. Mahajan, S. Tople, and A. Sharma, “Domain generalization using causal matching,” in International conference on machine learning . PMLR, 2021, pp. 7313–7324

    2021

  46. [54]

    Domain generalization via contrastive causal learning,

    Q. Miao, J. Yuan, and K. Kuang, “Domain generalization via contrastive causal learning,” arXiv preprint arXiv:2210.02655 , 2022

    2022 arXiv

  47. [55]

    Invariant causal representation learning for out-of-distribution generalization,

    C. Lu, Y . Wu, J. M. Hern ´andez-Lobato, and B. Sch ¨olkopf, “Invariant causal representation learning for out-of-distribution generalization,” in International Conference on Learning Representations , 2021

    2021

  48. [56]

    Causal transportability for visual recognition,

    C. Mao, K. Xia, J. Wang, H. Wang, J. Yang, E. Bareinboim, and C. V on- drick, “Causal transportability for visual recognition,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 7521–7531

    2022

  49. [57]

    Integrating markov blanket discovery into causal representation learning for domain generalization,

    N. Yin, H. Wang, Y . Yu, T. Gao, A. Dhurandhar, and Q. Ji, “Integrating markov blanket discovery into causal representation learning for domain generalization,” in European Conference on Computer Vision. Springer, 2024, pp. 271–288

    2024

  50. [58]

    Partial disentanglement for domain adaptation,

    L. Kong, S. Xie, W. Yao, Y . Zheng, G. Chen, P. Stojanov, V . Akinwande, and K. Zhang, “Partial disentanglement for domain adaptation,” in International conference on machine learning . PMLR, 2022, pp. 11 455–11 472

    2022

  51. [59]

    Show, attend and tell: neural image caption gen- eration with visual attention,

    K. Xu, J. L. Ba, R. Kiros, K. Cho, A. Courville, R. Salakhutdinov, R. S. Zemel, and Y . Bengio, “Show, attend and tell: neural image caption gen- eration with visual attention,” in Proceedings of the 32nd International Conference on International Conference on Machine Learning...

    2015

  52. [60]

    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

    2017

  53. [61]

    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,” in Proceedings of the IEEE international conference on computer vision , 2013, pp. 1657–1664

    2013

  54. [62]

    Deep hashing network for unsupervised domain adaptation,

    H. Venkateswara, J. Eusebio, S. Chakraborty, and S. Panchanathan, “Deep hashing network for unsupervised domain adaptation,” in Pro- ceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 5018–5027

    2017

  55. [63]

    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,” in Proceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 1406– 1415

    2019

  56. [64]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2016, pp. 770–778

    2016

  57. [65]

    An image is worth 16x16 words: Transformers for image recognition at scale,

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly et al., “An image is worth 16x16 words: Transformers for image recognition at scale,” in International Conference on Learning Representations

  58. [66]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,” in Proceedings of the 31st International Conference on Neural Information Processing Systems, 2017, pp. 6000–6010

    2017

  59. [67]

    Sgdr: Stochastic gradient descent with warm restarts,

    I. Loshchilov and F. Hutter, “Sgdr: Stochastic gradient descent with warm restarts,” arXiv preprint arXiv:1608.03983 , 2016

    2016 arXiv

  60. [68]

    Optimal representations for covariate shift,

    Y . Ruan, Y . Dubois, and C. J. Maddison, “Optimal representations for covariate shift,” in International Conference on Learning Representa- tions

  61. [69]

    Soft prompt generation for domain generalization,

    S. Bai, Y . Zhang, W. Zhou, Z. Luan, and B. Chen, “Soft prompt generation for domain generalization,” in European Conference on Computer Vision. Springer, 2024, pp. 434–450

    2024

  62. [70]

    Domain generalization by mutual-information regularization with pre-trained models,

    J. Cha, K. Lee, S. Park, and S. Chun, “Domain generalization by mutual-information regularization with pre-trained models,” European Conference on Computer Vision (ECCV) , 2022

    2022

  63. [71]

    Arcface: Additive angular margin loss for deep face recognition,

    J. Deng, J. Guo, N. Xue, and S. Zafeiriou, “Arcface: Additive angular margin loss for deep face recognition,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 4690– 4699

    2019

  64. [72]

    Visualizing data using t-sne,

    L. van der Maaten and G. Hinton, “Visualizing data using t-sne,” Journal of Machine Learning Research , vol. 9, pp. 2579–2605, 2008

    2008

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.