Pith. sign in

REVIEW 5 major objections 5 minor 50 references

Domain Adaptation Techniques for Natural and Medical Image Classification

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

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

Pith's one-line read Across 557 simulations on 13 datasets, this study finds domain adaptation strongly helps natural-image classification but yields little gain on most medical datasets, except for DSAN on COVID-19.

desk verdict A broad, useful DA benchmark, but the medical 'domain shift' experiments are mostly random splits of a single dataset, so the headline conclusion about medical DA is not supported. read the letter →

arxiv 2508.20537 v1 pith:SOC6DRL3 submitted 2025-08-28 cs.CV

Ahmad Chaddad , Yihang Wu , Reem Kateb , Christian Desrosiers This is my paper · ORCID
classification cs.CV
keywords domainadaptationimageclassificationmedicalimagingdeepsubdomainnetworkDSANout-of-distributiondynamicdatastreamsexplainableAI
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

This paper is a large comparative simulation study asking whether domain adaptation (DA), the practice of adapting a model trained on one data distribution so it also works on a different but related one, pays off for image classification in natural and medical settings. It runs seven DA techniques across 557 experiments on five natural and eight medical datasets, covering out-of-distribution inputs, limited training data, and dynamic data streams. The central finding is that DA substantially improves accuracy on natural-image benchmarks, while on most medical datasets the gains over a plain trained network are small or absent. The one clear exception is DSAN, the Deep Subdomain Adaptation Network, which reaches 91.2% accuracy on COVID-19 CT images with a ResNet-50 backbone and gains +6.7% over the baseline in the dynamic data-stream scenario. The authors also show that adding DA does not by itself guarantee interpretable or clinically meaningful attention, as measured by Grad-CAM.

What carries the argument

The machinery is the comparative DA pipeline rather than a single new algorithm. Each method is one of several standard image-classification networks (ResNet, DenseNet, ShuffleNet, MobileNet) trained with cross-entropy on the source domain plus a domain-alignment loss scaled by weight $\lambda$, and judged by highest test accuracy, by the A-distance (a classifier-based estimate of how far apart the two domains are), by t-SNE feature plots, and by Grad-CAM explanations. The method that carries the paper's best results is DSAN, the Deep Subdomain Adaptation Network, which replaces global maximum-mean-discrepancy alignment with local subdomain alignment: it weights source and target samples by class and matches each class separately in a kernel feature space. This class-wise local MMD (LMMD) is what the paper credits for DSAN's low A-distance, its 91.2% COVID-19 accuracy, and its +6.7% dynamic-stream gain.

What would settle it

Measure the A-distance between source and target on the random splits used in Section 3.4 (for instance, the 90/10 kidney-cancer split and the 10% cervical-cancer training sample); if that distance is near zero, those experiments do not exercise domain adaptation, and the decisive test is to rerun the same seven methods on cross-hospital or cross-scanner splits and compare the gains.

Watch

Extended reading notes

Core claim

The paper claims that DA is a mixed tool: it can greatly improve classification on computer-vision datasets, but for most medical imaging datasets the tested algorithms deliver only limited improvement over a baseline trained without adaptation. On the mainstream Office31 benchmark, DSAN reaches 88.4% average accuracy, a +12.3% gain over no DA; on the larger Adaptiope dataset, the best method (EUDA with a DINOv2 backbone) reaches 75.1%. On medical data, in contrast, most gains are marginal: DSAN's 91.2% on COVID-19 and strong performance on kidney cancer stand out, but on ChestXray8 and the multisource skin-cancer set DA often fails to beat the no-DA baseline; on kidney and cervical cancer the main benefit is reaching top accuracy earlier in training rather than exceeding it. The paper's headline result is that DSAN, which aligns class-wise subdomains rather than whole distributions, is the most consistently effective method, including a +6.7% accuracy improvement in the dynamic data-stream COVID-19 scenario, and that DSAN's Grad-CAM explanations are more localized on COVID-19 and skin-cancer images.

Load-bearing premise

The load-bearing premise, set up in Section 3.4.1, is that randomly splitting a single medical dataset into source and target portions creates a meaningful distribution shift; if those portions are nearly identical, the conclusion that DA gives little medical benefit is an artifact of the experimental design.

Editorial extensions

If this is right

  • On natural-image benchmarks, distribution-matching (MMD-based) methods such as DSAN and DCAN and adversarial methods such as DALN are the reliable choices; correlation-based Deep Coral is the weakest, especially on large-scale datasets.
  • On challenging medical datasets such as ChestXray8 and multisource skin cancer, plain training is often as good as or better than DA, suggesting that class imbalance or subtle lesions, not domain shift, are the real bottleneck.
  • In limited-sample and out-of-distribution settings, DA can still help: DALN and DSAN improve robustness to Gaussian blur, random erasing, and small training sets such as LCOVID-19.
  • In dynamic data streams, where the training distribution changes over time, DSAN and DALN recover several points of accuracy on COVID-19, showing DA can support continually updated medical classifiers.
  • Interpretability is not a free by-product of DA: Grad-CAM heatmaps show DSAN and BNM focusing on tighter regions, while the no-DA baseline can fixate on clinically irrelevant structures such as bone tissue.

Reading between the lines

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

  • The medical conclusions rest on random train/test splits of each dataset; a fairer test would use cross-site, cross-scanner, or cross-population splits, where a true domain shift exists and DA might show larger gains than reported here.
  • The dynamic-stream setup changes the training data each epoch, so DSAN's +6.7% gain may partly reflect optimization on fresh data rather than domain alignment; retraining on shuffled data would separate the two effects.
  • The paper's own closing limitation, that only cross-entropy loss and one hyperparameter grid were tested, means the observed medical ceiling could be tied to training choices rather than to DA itself.
  • For low-memory deployment, the batch-size results imply a practical rule: at batch size 4, MMD and nuclear-norm methods collapse while Deep Coral and DANN stay stable, so correlation or adversarial DA is the safer choice on edge devices.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 5 minor

Summary. This manuscript reports a large-scale empirical study of seven domain adaptation (DA) techniques (Deep Coral, DANN, DSAN, BNM, DALN, DCAN, and EUDA) applied to image classification in five natural and eight medical datasets. The authors evaluate accuracy, t-SNE visualizations, A-distance, and Grad-CAM interpretability, and they also probe settings with limited training samples, small batch sizes, out-of-distribution transformations, and a dynamic data-stream protocol. The main claims are that DA substantially improves natural-image classification, that DSAN performs outstandingly on medical data (e.g., 91.2% on COVID-19 and +6.7% in the dynamic stream scenario), and that DA yields only limited improvement on most medical imaging datasets.

Significance. The study's breadth is a genuine strength: it benchmarks seven methods across thirteen datasets, uses public data, reports multiple backbone architectures, and includes interpretability analyses. If the conclusions were supported by the experimental design, the paper would be a useful reference for practitioners choosing DA methods for medical imaging. However, the central medical-domain conclusion is undermined by the experimental setup: in Section 3.4.1, the 'source' and 'target' domains for most medical datasets are random or provided splits of a single dataset, so they are approximately identically distributed and do not constitute a domain shift. In addition, all accuracy tables report the best accuracy reached at any epoch without standard deviations or multiple seeds, which makes reported differences such as 91.2% versus 90.5% uninterpretable. The natural-image experiments (Office31, Adaptiope, MultiImageNet, cross-dataset) do involve genuine domain shifts and are more informative, but the paper's headline conclusions about medical DA cannot be drawn from the present evidence.

major comments (5)
  1. [Section 3.4.1] The medical 'domain shift' is degenerate for most datasets. Kidney cancer is a random 90/10 split; cervical cancer is a random 10% class-balanced training sample; COVID-19, ChestXray8, skin cancer, and retinal disease use the provided or random train/test partitions of a single dataset. These source and target sets are approximately i.i.d. samples from the same distribution, so there is no distribution shift for DA to correct. The conclusion in Section 6 that DA provides limited improvement on most medical datasets is therefore an artifact of the experimental design rather than a property of the methods. This also contradicts the Section 1 statement that the source and target domains have 'significant data distributions'; the authors should either use genuine cross-domain medical data (e.g., different scanners, hospitals, or modalities) or restrict the medical claims to the multisource skin cancer setting, which at least involves distinct source datasets.
  2. [Tables 3–14 and Algorithm 1] All accuracy tables report the highest accuracy achieved at any training epoch, which is effectively model selection on the test set, and no standard deviations or multiple seeds are reported. This inflates performance and makes small differences (e.g., DSAN 91.2% vs. w/o DA 90.5% on COVID-19) statistically uninterpretable. The paper itself cites [50] to warn about test-set-based model selection, yet the evaluation protocol does exactly this. The authors should report validation-based model selection or fixed-epoch evaluation, and include mean ± std over at least three seeds.
  3. [Section 4.5, Figure 14] The A-distance plots are measured on the same degenerate source/target splits as the accuracy results, so they cannot establish that DA provides meaningful alignment for medical data. Moreover, DSAN explicitly minimizes a local MMD loss, so a lower A-distance for DSAN is by construction and does not independently demonstrate that the alignment is beneficial for classification. The A-distance evidence should be either computed on genuine target domains or removed as support for the medical conclusions.
  4. [Section 4.6, Table 14] The dynamic data-stream protocol does not create a distribution shift over time. Dividing the Adaptiope source domain P into 30 non-overlapping parts and the COVID-19 training set into 100 non-overlapping parts, then training sequentially on these parts, produces source subsets that are i.i.d. draws from the same source distribution; the target distribution is fixed. The +6.7% improvement attributed to DSAN in the dynamic stream scenario therefore measures sequential training on random partitions, not adaptation to a changing target distribution. A meaningful dynamic DA evaluation requires a target stream that evolves over time (e.g., time-ordered data or a shifting corruption process).
  5. [Table 8 and Section 3.4.2 (Multisource skin cancer)] In Table 8, DSAN and w/o DA report numerically identical values for all three backbones and all three metrics (ACC, BACC, F1), e.g., 88.7/83.7/83.8 for Resnet34, 90.4/85.9/86.8 for Resnet50, and 89.3/84.8/85.1 for Densenet121. This is either a copy-paste error or indicates that the DSAN adaptation term had no effect; in either case it undermines the reliability of the table and the surrounding discussion. The authors must correct this and verify the reported values.
minor comments (5)
  1. [Abstract and Introduction] The abstract and introduction state '557 simulation studies' but the counting method is not explained; please clarify how the number of experiments is derived from the methods, datasets, backbones, and settings.
  2. [Section 3.3.3] The text says 'The hyperparameters settings are listed in Table 2,' but Table 2 is the dataset description table; the hyperparameters are in Table 1. Please correct the cross-reference.
  3. [Equation (8)] In the definition of the EUDA MMD loss, the notation '|D_s|t|' appears to contain a typo; it should likely be '|D_s|' and '|D_t|'.
  4. [Table 3 (Adaptiope)] The table includes a row for 'SACAEM' but this method is not described in Section 2 and is not listed among the seven algorithms in the introduction; please either describe it or remove it.
  5. [Figure 11 caption] The caption says the columns represent Resnet34, Resnet50, and Densenet121, but the experiment in Section 4.1 uses Resnet50 only; the caption should be corrected to match the experiment.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper is an empirical benchmark study with no derivation of predictions from fitted inputs.

full rationale

This manuscript reports a simulation/benchmark study rather than a derivation chain. The methods (Deep Coral, DANN, DSAN, BNM, DALN, DCAN, EUDA) are published external algorithms applied to external datasets; no parameter is fitted to a subset of data and then renamed as a prediction. The headline DSAN figures (91.2% on COVID-19; +6.7% in dynamic stream) are measured accuracies, not quantities implied by the algorithm's definition. The paper's self-citation [2] (Chaddad et al.) appears in the introduction as background for the DA concept and is not load-bearing for any reported result. The A-distance observation in Section 4.5 that DSAN reduces source-target distance is consistent with DSAN's loss being a subdomain MMD, but the paper reports it as an observational check, not as an independent prediction, so it is not circular in the sense of Eq. X = Eq. Y by construction. The most serious issue—the medical domain splits in Section 3.4.1 being random train/test partitions of the same dataset—is a threat to experimental validity and to the generalization claim about medical DA, but it is not a circularity of derivation; the paper's stated limitations concern hyperparameters and test-set reuse, not circular reasoning. Hence no circular step can be exhibited with the required specificity.

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

The central claim is empirical, so the ledger lists the hand-chosen hyperparameters and the domain assumptions needed to interpret the results. No new entities are introduced.

free parameters (4)
  • DA loss weight lambda = DC:10, DANN:1, DSAN:0.5, BNM:1, DALN:0.1, DCAN:0.5, EUDA:0.3
    Set by hand in Table 1 for each algorithm; no sensitivity analysis, yet it directly controls the balance between classification and adaptation and thus the reported accuracies.
  • Learning rate = 3e-3 to 3e-2 depending on algorithm
    Chosen per algorithm in Table 1; affects convergence and final best-epoch accuracy.
  • Batch size = 16 for baseline, 8 and 4 in stress tests
    Chosen by the authors; Section 4.1 shows results vary strongly with batch size, indicating the comparison is sensitive to this choice.
  • Iterations per epoch = 200 for most, 300 for COVID-19, 1000 for ChestXray8
    Ad hoc per dataset (Table 1 and Section 3.4.2); affects training length and best-epoch values.
assumptions (4)
  • domain assumption Random train/test splits of a single dataset create meaningful source and target domains
    Used to construct DA tasks for kidney, cervical, COVID-19, and other medical data (Section 3.4.1). If the splits are i.i.d., the benchmark is degenerate.
  • domain assumption Aligning feature distributions between source and target improves target classification
    The shared premise of all tested DA methods; not independently verified here.
  • domain assumption ImageNet-pretrained ResNet and DenseNet backbones provide a suitable feature space for medical images
    All methods use these backbones (Section 3.1); no domain-specific pretraining is tested.
  • domain assumption Grad-CAM heatmaps reflect clinically meaningful model attention
    Used in Section 4.7 to judge interpretability; a known approximation, not a verified explanation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Domain Adaptation Techniques for Natural and Medical Image Classification." pith.science (2026). https://pith.science/paper/SOC6DRL3

@misc{pith2026250820537,
  author       = {Pith},
  title        = {Pith review of: Domain Adaptation Techniques for Natural and Medical Image Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SOC6DRL3}},
  note         = {Machine review of arXiv:2508.20537}
}
read the original abstract

Domain adaptation (DA) techniques have the potential in machine learning to alleviate distribution differences between training and test sets by leveraging information from source domains. In image classification, most advances in DA have been made using natural images rather than medical data, which are harder to work with. Moreover, even for natural images, the use of mainstream datasets can lead to performance bias. {With the aim of better understanding the benefits of DA for both natural and medical images, this study performs 557 simulation studies using seven widely-used DA techniques for image classification in five natural and eight medical datasets that cover various scenarios, such as out-of-distribution, dynamic data streams, and limited training samples.} Our experiments yield detailed results and insightful observations highlighting the performance and medical applicability of these techniques. Notably, our results have shown the outstanding performance of the Deep Subdomain Adaptation Network (DSAN) algorithm. This algorithm achieved feasible classification accuracy (91.2\%) in the COVID-19 dataset using Resnet50 and showed an important accuracy improvement in the dynamic data stream DA scenario (+6.7\%) compared to the baseline. Our results also demonstrate that DSAN exhibits remarkable level of explainability when evaluated on COVID-19 and skin cancer datasets. These results contribute to the understanding of DA techniques and offer valuable insight into the effective adaptation of models to medical data.

Figures

Figures reproduced from arXiv: 2508.20537 by the authors.

Figure 1
Figure 1. Examples of domain adaptation (DA). (Left): There exists a large gap between source and target domain data distributions. (Right): After applying DA, the data distribution discrepancies decreased, and the target domain data distribution is more fit with the source data. The color circles represent different classes. MMD-based (i.e., measure and minimize the distribution differences between source and target domains … view at source ↗
Figure 2
Figure 2. summarizes the search results for DA-related articles that have been published in leading journals or conferences, with the results obtained from the Google Scholar and PubMed databases for the period between 2016 and 2022. As shown, this field has been the focus of a remarkable number of papers in recent years [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Example of samples from each dataset. 3.2. Evaluation metrics The methods’ effectiveness was assessed by observing the change in classification accuracy before and after the use of DA. Given the number of total samples 𝑁, the model’s prediction 𝑦𝑝𝑟𝑒𝑑, ground-truth label 𝑦𝑡𝑟𝑢𝑒, the classification accuracy can be defined as: 𝐴𝑐𝑐𝑢𝑟𝑎𝑐𝑦 = 1 𝑁 ∑ 𝑁 𝑖=1 1 𝑦 𝑝𝑟𝑒𝑑 𝑖 =𝑦 𝑡𝑟𝑢𝑒 𝑖 (14) Ahmad Chaddad et al.: Preprint submitted to E… view at source ↗
Figures from the paper (14 more)
Figure 4
Figure 4. Figure 4: The simulation pipeline and its stages. The training process for backpropagation involves the computation of the domain adaptation loss between the source and target domains at the bottleneck layer, along with the cross-entropy loss on the source domain. We also compar…
Figure 5
Figure 5. Figure 5: The visualization examples of the learned representations using t-SNE for the DA methods using MultiImageNet dataset with Resnet34 as backbone. The red color represents source samples, while the blue color represents target samples [PITH_FULL_IMAGE:figures/full_fig_p0…
Figure 6
Figure 6. Figure 6: Test classification accuracy during training of the DA methods in the cross-dataset with Resnet50 as backbone. The x-axis represents the number of training epochs, while the y-axis shows the classification accuracy [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: Test classification accuracy obtained using six DA techniques on medical datasets. The x-axis represents the number of training epochs. The columns, from left to right, represent Resnet34, Resnet50, and Densenet121, respectively. Kidney cancer The parameters used for t…
Figure 8
Figure 8. Figure 8: The visualization examples of the learned representations using t-SNE for the DA methods using Skin cancer dataset with Resnet50 as backbone. The red color represents source samples, while the blue color represents target samples. 8 gives a t-SNE plot of the representa…
Figure 9
Figure 9. Figure 9: The visualization examples of the learned representations using t-SNE for the DA methods using multisource skin cancer dataset with Resnet50 as backbone. The red color represents source samples, while the blue color represents target samples. Densenet121 alone). This s…
Figure 10
Figure 10. Figure 10: Test classification accuracy obtained using six DA techniques on medical datasets. The x-axis represents the number of training epochs. The columns, from left to right, represent Resnet34, Resnet50, and Densenet121, respectively. results suggest that LMMD is less effe…
Figure 11
Figure 11. Figure 11: Test classification accuracy using six DA techniques on COVID-19 dataset with three batch sizes. The columns, from left to right, represent Resnet34, Resnet50, and Densenet121, respectively. The x-axis represents the number of training epochs. the DALN algorithm on Sh…
Figure 12
Figure 12. Figure 12: Test classification accuracy on COVID-19 dataset for out-of-distribution scenario. The columns, from left to right, represent Resnet34, Resnet50, and Densenet121, respectively. The x-axis represents the number of training epochs [PITH_FULL_IMAGE:figures/full_fig_p020…
Figure 13
Figure 13. Figure 13: Test classification accuracy using six DA techniques on LCOVID-19 dataset. The columns, from left to right, represent Resnet34, Resnet50, and Densenet121, respectively. 0 5 10 15 20 25 30 4 8 12 16 20 -distance Skin cancer DeepCoral DANN DSAN BNM w/oDA (Resnet50) DALN…
Figure 14
Figure 14. Figure 14: -distance for the DA techniques using four datasets (i.e., skin cancer, retinal disease, kidney and cervical cancer). The Y-axis represents the -distance value, whereas the X-axis represents the training epochs. The -distance is a metric commonly used to estimate t…
Figure 15
Figure 15. Figure 15: The left column represents the classification accuracy achieved with six DA techniques on the Adaptiope and COVID-19 datasets, with the x-axis indicating the training epochs. The right column shows the sample number in each data stream. COVID-19 w/o DA Deep Coral DANN…
Figure 16
Figure 16. Figure 16: Example of heatmap visualization using Grad-CAM. The model’s attention is more focused on its current location when the color is deeper, such as red. class features by generating high-quality samples. The results obtained for different network backbones (Resnet34, Res…
Figure 17
Figure 17. Figure 17: Example of texture and style shifts in MultiImageNet dataset. Orange circle indicates the specific class of this image. ImageNet-R shows wide shifts where only small parts represent the class specific object. Furthermore, the implementation of stringent data privacy p…

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

50 extracted references · 41 canonical work pages

  1. [50]

    DoCIFAR-10classifiersgeneralizetoCIFAR-10? arXivpreprint arXiv:1806.00451, 2018

    BenjaminRecht,RebeccaRoelofs,LudwigSchmidt,andVaishaalShankar. DoCIFAR-10classifiersgeneralizetoCIFAR-10? arXivpreprint arXiv:1806.00451, 2018. Ahmad Chaddad et al.:Preprint submitted to Elsevier Page 27 of 27

    arXiv 2018

  2. [1]

    Unpaired,unsuperviseddomainadaptationassumesyourdomainsarealreadysimilar

    GijsvanTulderandMarleendeBruijne. Unpaired,unsuperviseddomainadaptationassumesyourdomainsarealreadysimilar. MedicalImage Analysis, 87:102825, 2023

    work page 2023

  3. [2]

    Simulations of common unsupervised domain adaptation algorithms for image classification.IEEE Transactions on Instrumentation and Measurement, 74:1–17, 2025

    Ahmad Chaddad, Yihang Wu, Yuchen Jiang, Ahmed Bouridane, and Christian Desrosiers. Simulations of common unsupervised domain adaptation algorithms for image classification.IEEE Transactions on Instrumentation and Measurement, 74:1–17, 2025

    work page 2025

  4. [3]

    Domain adaptation in remote sensing image classification: A survey

    Jiangtao Peng, Yi Huang, Weiwei Sun, Na Chen, Yujie Ning, and Qian Du. Domain adaptation in remote sensing image classification: A survey. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 15:9842–9859, 2022

    work page 2022

  5. [4]

    Correlation alignment for unsupervised domain adaptation.Domain adaptation in computer vision applications, pages 153–171, 2017

    Baochen Sun, Jiashi Feng, and Kate Saenko. Correlation alignment for unsupervised domain adaptation.Domain adaptation in computer vision applications, pages 153–171, 2017

    work page 2017

  6. [5]

    Deep subdomain adaptation network for image classification.IEEE transactions on neural networks and learning systems, 32(4):1713–1722, 2020

    Yongchun Zhu, Fuzhen Zhuang, Jindong Wang, Guolin Ke, Jingwu Chen, Jiang Bian, Hui Xiong, and Qing He. Deep subdomain adaptation network for image classification.IEEE transactions on neural networks and learning systems, 32(4):1713–1722, 2020

    work page 2020

  7. [6]

    Unsuperviseddomainadaptationviadeepconditionaladaptationnetwork

    PengfeiGe,Chuan-XianRen,Xiao-LinXu,andHongYan. Unsuperviseddomainadaptationviadeepconditionaladaptationnetwork. Pattern Recognition, 134:109088, 2023

    work page 2023

  8. [7]

    Deep CORAL: Correlation alignment for deep domain adaptation

    Baochen Sun and Kate Saenko. Deep CORAL: Correlation alignment for deep domain adaptation. InEuropean conference on computer vision, pages 443–450. Springer, 2016

    work page 2016

Show all 50 references
  1. [8]

    Heterogeneousdomainadaptationviacorrelativeanddiscriminativefeature learning

    YuwuLu,DeweiLin,LinlinShen,YicongZhou,andJiahuiPan. Heterogeneousdomainadaptationviacorrelativeanddiscriminativefeature learning. IEEE Transactions on Multimedia, 27:3447–3461, 2025

    2025

  2. [9]

    Weakly correlated multimodal domain adaptation for pattern classification

    Shuyue Wang, Zhunga Liu, Zuowei Zhang, and Mohammed Bennamoun. Weakly correlated multimodal domain adaptation for pattern classification. IEEE Transactions on Artificial Intelligence, 6(5):1360–1372, 2025

    2025

  3. [10]

    Domain-adversarial training of neural networks.The journal of machine learning research, 17(1):2096–2030, 2016

    Yaroslav Ganin, Evgeniya Ustinova, Hana Ajakan, Pascal Germain, Hugo Larochelle, François Laviolette, Mario Marchand, and Victor Lempitsky. Domain-adversarial training of neural networks.The journal of machine learning research, 17(1):2096–2030, 2016

    2016

  4. [11]

    Reusing the task-specific classifier as a discriminator: Discriminator-free adversarial domain adaptation

    Lin Chen, Huaian Chen, Zhixiang Wei, Xin Jin, Xiao Tan, Yi Jin, and Enhong Chen. Reusing the task-specific classifier as a discriminator: Discriminator-free adversarial domain adaptation. In2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 7171–...

    2022

  5. [12]

    InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 3941–3950, 2020

    ShuhaoCui,ShuhuiWang,JunbaoZhuo,LiangLi,QingmingHuang,andQiTian.Towardsdiscriminabilityanddiversity:Batchnuclear-norm maximization under label insufficient situations. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 3941–3950, 2020

    2020

  6. [13]

    Faa-clip:Federatedadversarialadaptationofclip

    YihangWu,AhmadChaddad,ChristianDesrosiers,TareefDaqqaq,andReemKateb. Faa-clip:Federatedadversarialadaptationofclip. IEEE Internet of Things Journal, 12(12):21091–21102, 2025

    2025

  7. [14]

    Domain-guided conditional diffusion model for unsupervised domain adaptation.Neural Networks, 184:107031, 2025

    Yulong Zhang, Shuhao Chen, Weisen Jiang, Yu Zhang, Jiangang Lu, and James T Kwok. Domain-guided conditional diffusion model for unsupervised domain adaptation.Neural Networks, 184:107031, 2025

    2025

  8. [15]

    Multi-source domain adaptation by causal-guided adaptive multimodal diffusion networks.International Journal of Computer Vision, 133(7):4623–4645, 2025

    Ziyun Cai, Yawen Huang, Tengfei Zhang, Yefeng Zheng, and Dong Yue. Multi-source domain adaptation by causal-guided adaptive multimodal diffusion networks.International Journal of Computer Vision, 133(7):4623–4645, 2025

    2025

  9. [16]

    Ddci: Unsupervised domain adaptation for remote sensing images based on diffusion causal distillation.IEEE Transactions on Geoscience and Remote Sensing, 63:1–12, 2025

    Jiaqi Zhao, Yao Li, Yong Zhou, Wen-Liang Du, Xixi Li, Rui Yao, and Abdulmotaleb El Saddik. Ddci: Unsupervised domain adaptation for remote sensing images based on diffusion causal distillation.IEEE Transactions on Geoscience and Remote Sensing, 63:1–12, 2025

    2025

  10. [17]

    Textadapter: Self-supervised domain adaptation for cross-domain text recognition.IEEE Transactions on Multimedia, 26:9854–9865, 2024

    Xiao-Qian Liu, Peng-Fei Zhang, Xin Luo, Zi Huang, and Xin-Shun Xu. Textadapter: Self-supervised domain adaptation for cross-domain text recognition.IEEE Transactions on Multimedia, 26:9854–9865, 2024

    2024

  11. [18]

    Samda: Leveraging sam on few-shot domain adaptation for electronic microscopy segmentation

    Yiran Wang and Li Xiao. Samda: Leveraging sam on few-shot domain adaptation for electronic microscopy segmentation. In2025 IEEE 22nd International Symposium on Biomedical Imaging (ISBI), pages 1–5. IEEE, 2025

    2025

  12. [19]

    Feature fusion transferability aware transformer for unsupervised domain adaptation

    Xiaowei Yu, Zhe Huang, and Zao Zhang. Feature fusion transferability aware transformer for unsupervised domain adaptation. InWinter Conference on Applications of Computer Vision (WACV), pages 6752–6761, 2025

    2025

  13. [20]

    Robust unsupervised domain adaptation through negative-view regularization

    Joonhyeok Jang, Sunhyeok Lee, Seonghak Kim, Jung-Un Kim, Seonghyun Kim, and Daeshik Kim. Robust unsupervised domain adaptation through negative-view regularization. InProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 2462– 2471, 2024

    2024

  14. [21]

    Open-set domain adaptation with visual-language foundation models.Computer Vision and Image Understanding, 250:104230, 2025

    Qing Yu, Go Irie, and Kiyoharu Aizawa. Open-set domain adaptation with visual-language foundation models.Computer Vision and Image Understanding, 250:104230, 2025

    2025

  15. [22]

    Adapting visual category models to new domains

    Kate Saenko, Brian Kulis, Mario Fritz, and Trevor Darrell. Adapting visual category models to new domains. InEuropean Conference on Computer Vision, pages 213–226. Springer, 2010

    2010

  16. [23]

    Reusing the task-specific classifier as a discriminator: Discriminator-freeadversarialdomainadaptation

    Lin Chen, Huaian Chen, Zhixiang Wei, Xin Jin, Xiao Tan, Yi Jin, and Enhong Chen. Reusing the task-specific classifier as a discriminator: Discriminator-freeadversarialdomainadaptation. In ProceedingsoftheIEEE/CVFConferenceonComputerVisionandPatternRecognition (CVPR), pages 718...

    2022

  17. [24]

    A brief review of domain adaptation.Advances in data science and information engineering, pages 877–894, 2021

    Abolfazl Farahani, Sahar Voghoei, Khaled Rasheed, and Hamid R Arabnia. A brief review of domain adaptation.Advances in data science and information engineering, pages 877–894, 2021

    2021

  18. [25]

    NeuralNetworks, page 106230, 2024

    YuqiFang,Pew-ThianYap,WeiliLin,HongtuZhu,andMingxiaLiu.Source-freeunsuperviseddomainadaptation:Asurvey. NeuralNetworks, page 106230, 2024

    2024

  19. [26]

    Euda: An efficient unsupervised domain adaptation via self-supervised vision transformer

    Ali Abedi, QM Wu, Ning Zhang, and Farhad Pourpanah. Euda: An efficient unsupervised domain adaptation via self-supervised vision transformer. arXiv preprint arXiv:2407.21311, 2024

    2024 arXiv

  20. [27]

    Large-scale machine learning with stochastic gradient descent

    Léon Bottou. Large-scale machine learning with stochastic gradient descent. In Proceedings of COMPSTAT’2010: 19th International ConferenceonComputationalStatisticsParisFrance,August22-27,2010Keynote,InvitedandContributedPapers ,pages177–186.Springer, 2010

    2010

  21. [28]

    Visualizing data using t-SNE.Journal of machine learning research, 9(11), 2008

    Laurens Van der Maaten and Geoffrey Hinton. Visualizing data using t-SNE.Journal of machine learning research, 9(11), 2008

    2008

  22. [29]

    Analysis of representations for domain adaptation.Advances in neural information processing systems, 19, 2006

    Shai Ben-David, John Blitzer, Koby Crammer, and Fernando Pereira. Analysis of representations for domain adaptation.Advances in neural information processing systems, 19, 2006

    2006

  23. [30]

    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. InProceedings of the IEEE international conference on computer vision, pages 618–626, 2017

    2017

  24. [31]

    Deep hashing network for unsupervised domain adaptation

    Hemanth Venkateswara, Jose Eusebio, Shayok Chakraborty, and Sethuraman Panchanathan. Deep hashing network for unsupervised domain adaptation. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 5018–5027, 2017

    2017

  25. [32]

    Themanyfacesofrobustness:Acriticalanalysisofout-of-distributiongeneralization

    Dan Hendrycks, Steven Basart, Norman Mu, Saurav Kadavath, Frank Wang, Evan Dorundo, Rahul Desai, Tyler Zhu, Samyak Parajuli, Mike Guo,DawnSong,JacobSteinhardt,andJustinGilmer. Themanyfacesofrobustness:Acriticalanalysisofout-of-distributiongeneralization. ICCV, 2021

    2021

  26. [33]

    Learning robust global representations by penalizing local predictive power

    Haohan Wang, Songwei Ge, Zachary Lipton, and Eric P Xing. Learning robust global representations by penalizing local predictive power. In Advances in Neural Information Processing Systems, pages 10506–10518, 2019

    2019

  27. [34]

    Adaptiope: A modern benchmark for unsupervised domain adaptation

    Tobias Ringwald and Rainer Stiefelhagen. Adaptiope: A modern benchmark for unsupervised domain adaptation. In2021 IEEE Winter Conference on Applications of Computer Vision (WACV), pages 101–110, 2021

    2021

  28. [35]

    COVIDNet-CT: A tailored deep convolutional neural network design for detection of COVID-19 cases from chest CT images.Frontiers in Medicine, 7:1025, 2020

    Hayden Gunraj, Linda Wang, and Alexander Wong. COVIDNet-CT: A tailored deep convolutional neural network design for detection of COVID-19 cases from chest CT images.Frontiers in Medicine, 7:1025, 2020

    2020

  29. [36]

    ChestX-ray8: Hospital-scale chest X-raydatabaseandbenchmarksonweakly-supervisedclassificationandlocalizationofcommonthoraxdiseases

    Xiaosong Wang, Yifan Peng, Le Lu, Zhiyong Lu, Mohammadhadi Bagheri, and Ronald M Summers. ChestX-ray8: Hospital-scale chest X-raydatabaseandbenchmarksonweakly-supervisedclassificationandlocalizationofcommonthoraxdiseases. In ProceedingsoftheIEEE conference on computer vision a...

    2017

  30. [37]

    Vision transformer and explainabletransferlearningmodelsforautodetectionofkidneycyst,stoneandtumorfromCT-radiography

    Md Nazmul Islam, Mehedi Hasan, Md Kabir Hossain, Md Golam Rabiul Alam, Md Zia Uddin, and Ahmet Soylu. Vision transformer and explainabletransferlearningmodelsforautodetectionofkidneycyst,stoneandtumorfromCT-radiography. ScientificReports,12(1):11440, 2022

    2022

  31. [38]

    Retinal fundus multi-disease image dataset (RFMiD): A dataset for multi-disease detection research

    Samiksha Pachade, Prasanna Porwal, Dhanshree Thulkar, Manesh Kokare, Girish Deshmukh, Vivek Sahasrabuddhe, Luca Giancardo, Gwenolé Quellec, and Fabrice Mériaudeau. Retinal fundus multi-disease image dataset (RFMiD): A dataset for multi-disease detection research. Data, 6(2):14, 2021

    2021

  32. [39]

    The ham10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions.Scientific data, 5(1):1–9, 2018

    Philipp Tschandl, Cliff Rosendahl, and Harald Kittler. The ham10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions.Scientific data, 5(1):1–9, 2018

    2018

  33. [40]

    NoelCFCodella,DavidGutman,MEmreCelebi,BrianHelba,MichaelAMarchetti,StephenWDusza,AadiKalloo,KonstantinosLiopyris, Nabin Mishra, Harald Kittler, et al. Skin lesion analysis toward melanoma detection: A challenge at the 2017 international symposium on biomedical imaging (isbi), ...

    2017

  34. [41]

    Bcn20000: Dermoscopic lesions in the wild.arXiv preprint arXiv:1908.02288, 2019

    Marc Combalia, Noel CF Codella, Veronica Rotemberg, Brian Helba, Veronica Vilaplana, Ofer Reiter, Cristina Carrera, Alicia Barreiro, Allan C Halpern, Susana Puig, et al. Bcn20000: Dermoscopic lesions in the wild.arXiv preprint arXiv:1908.02288, 2019

    1908 arXiv

  35. [42]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016

    2016

  36. [43]

    Densely connected convolutional networks

    Gao Huang, Zhuang Liu, Laurens Van Der Maaten, and Kilian Q Weinberger. Densely connected convolutional networks. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 4700–4708, 2017

    2017

  37. [44]

    Shufflenet:Anextremelyefficientconvolutionalneuralnetworkformobiledevices

    XiangyuZhang,XinyuZhou,MengxiaoLin,andJianSun. Shufflenet:Anextremelyefficientconvolutionalneuralnetworkformobiledevices. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 6848–6856, 2018

    2018

  38. [45]

    MobileNetV2: Inverted residuals and linear bottlenecks

    Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, and Liang-Chieh Chen. MobileNetV2: Inverted residuals and linear bottlenecks. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 4510–4520, 2018

    2018

  39. [46]

    Multi-source and multi-target domain adaptation based on dynamic generator with attention.IEEE Transactions on Multimedia, 26:6891–6905, 2024

    Yuwu Lu, Haoyu Huang, Biqing Zeng, Zhihui Lai, and Xuelong Li. Multi-source and multi-target domain adaptation based on dynamic generator with attention.IEEE Transactions on Multimedia, 26:6891–6905, 2024

    2024

  40. [47]

    Federated learning for healthcare applications.IEEE Internet of Things Journal, 11(5):7339–7358, 2024

    Ahmad Chaddad, Yihang Wu, and Christian Desrosiers. Federated learning for healthcare applications.IEEE Internet of Things Journal, 11(5):7339–7358, 2024

    2024

  41. [48]

    Adversarial domain adaptation with clip for few-shot image classification

    Tongfeng Sun, Hongjian Yang, Zhongnian Li, Xinzheng Xu, and Xiurui Wang. Adversarial domain adaptation with clip for few-shot image classification. Applied Intelligence, 55(1):59, 2025

    2025

  42. [49]

    Calibratingdeepneuralnetworksusing focal loss.Advances in Neural Information Processing Systems, 33:15288–15299, 2020

    JishnuMukhoti,VivekaKulharia,AmartyaSanyal,StuartGolodetz,PhilipTorr,andPuneetDokania. Calibratingdeepneuralnetworksusing focal loss.Advances in Neural Information Processing Systems, 33:15288–15299, 2020

    2020

Pith tools

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