Efficient Few-Shot Medical Image Analysis via Hierarchical Contrastive Vision-Language Learning

Fernando Gabriela Garcia; Harrison Fuller; Victor Flores

REVIEW 4 major objections 4 minor 22 references

Efficient Few-Shot Medical Image Analysis via Hierarchical Contrastive Vision-Language Learning

T0 review · 4 major / 4 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read HiCA's two-stage, three-level contrastive alignment yields the best few-shot accuracy and AUC on two medical imaging benchmarks.

desk verdict A coherent method paper whose central SOTA claim is unverifiable because the key baseline and all uncertainty estimates are missing. read the letter →

arxiv 2501.09294 v1 pith:L76RCWVO submitted 2025-01-16 cs.CV cs.CL

Harrison Fuller , Fernando Gabriela Garcia , Victor Flores This is my paper · ORCID

classification cs.CVcs.CL

keywords few-shotlearningmedicalimageclassificationvision-languagemodelscontrastivehierarchicalalignmentchestX-raybreastultrasounddomainadaptation

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

The paper proposes HiCA, a two-stage fine-tuning strategy that adapts a large vision-language model to medical image classification using very few labeled examples. Stage one adapts the visual and text encoders to medical data using pseudo-labeled images and a medical text corpus; stage two aligns images and text at three levels: global image-text, local region-text, and cross-category separation. The paper reports that HiCA outperforms CLIP zero-shot, a supervised CNN, and ImageNet-transfer learning on Chest X-ray and Breast Ultrasound, reaching an AUC of 0.92 in the 20-shot breast ultrasound setting. If this holds, pretrained vision-language models become practical for medical domains where annotated data is scarce.

What carries the argument

The load-bearing object is the hierarchical contrastive loss itself: three additive terms that jointly enforce whole-image alignment, fine-grained region-level alignment, and separation between categories. The global term is the standard image-text contrastive loss, the local term repeats it on regions of interest extracted by a segmentation model, and the cross-category term is a margin loss on mismatched pairs. The ablation study is what makes this machinery central: removing any one term lowers Chest X-ray accuracy from 86.3 to between 83.5 and 84.7, so the paper credits the hierarchy, rather than any single loss, with the improvement.

What would settle it

Measure the segmentation masks used for $L_{\text{local}}$ against expert segmentations on a held-out subset of each dataset. If the ROI masks have low overlap with anatomical structures yet the reported AUC still holds, the local-loss gains are not caused by anatomical alignment. Conversely, retrain HiCA with only $L_{\text{global}} + L_{\text{cross}}$; if Breast Ultrasound AUC does not drop substantially, the central contribution of the local alignment term is unsupported.

Watch

Extended reading notes

Core claim

HiCA's central claim is that the gap between pretrained vision-language models and medical images is largely an alignment problem, and that aligning image and text embeddings at three levels closes it. The method fine-tunes the visual encoder on pseudo-labeled medical images and the text encoder on medical text, then optimizes $L_{\text{HiCA}} = L_{\text{global}} + \lambda_1 L_{\text{local}} + \lambda_2 L_{\text{cross}}$, where the global term matches whole images to class descriptors, the local term matches segmented regions of interest to fine-grained descriptors, and the cross-category term pushes mismatched pairs apart by a margin. The paper reports that this combination raises accuracy and AUC above CLIP zero-shot, supervised CNN, and ImageNet-transfer baselines on both datasets, with the largest gain an AUC of 0.92 in the 20-shot Breast Ultrasound setting.

Load-bearing premise

The method's success rests on an unnamed segmentation model that extracts the regions of interest; if that model is inaccurate or was trained on the evaluation datasets, the reported local-alignment gains could be artifacts rather than a property of hierarchical contrastive learning.

Editorial extensions

If this is right

On the two tested datasets, HiCA beats CLIP zero-shot, a supervised CNN, and ImageNet-transfer learning in both few-shot and zero-shot accuracy and AUC.
Ablations show that the global, local, and cross-category losses each contribute; dropping any one lowers Chest X-ray accuracy from 86.3 to 83.5-84.7.
On held-out unseen categories, HiCA generalizes better than all baselines, reaching 75.6 accuracy versus 71.8 for the best baseline.
Under noisy textual descriptors, HiCA degrades gracefully, losing 4.2 accuracy points on Chest X-ray compared with clean text.
In the paper's human evaluation with three radiologists, HiCA predictions scored higher than transfer learning on both interpretability and clinical validity.

Reading between the lines

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

My inference: if the local-alignment term works because ROIs capture clinically meaningful structure, the same loss should transfer to histopathology or retinal imaging, where fine-grained regions are similarly diagnostic.
My inference: because Stage 1 pseudo-labels come from unsupervised clustering, the method's ceiling likely depends on cluster quality; varying the cluster count or using a different clustering algorithm would expose that sensitivity.
My inference: the descriptor-generation process is underspecified, so a natural test is to replace the hand-written descriptors with LLM-generated reports and measure whether the noisy-descriptor robustness persists.

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.

Desk Editor's Note

read the letter

Quick take: this is a coherent combination of known pieces, but the central claim is not checkable. Table I omits Byra et al. [5], the very baseline the introduction says HiCA surpasses, and Section IV-A promises significance testing that never appears in any table. The stress-test note lands.

What's actually new: the package. Global CLIP-style alignment, ROI-level local alignment with a margin-based cross-category loss, plus the two-stage pseudo-label pretraining, is a sensible composition for medical few-shot VLMs. Equations (3)-(5) are clean, and the ablation, human eval, noisy-text robustness check, and efficiency table show real effort. I don't doubt the authors know what they want to do.

Soft spots, in order of severity. First, the SOTA claim is structurally unsupported: no [5], no other recent few-shot medical VLM, no error bars, no standard deviations, no p-values or confidence intervals anywhere in Tables I-VI. 'Consistently outperforms' is asserted, not demonstrated. Second, reproducibility: the ROI extraction uses an unnamed segmentation model, the clustering algorithm is not described, and lambda1, lambda2, tau, delta, and K are not reported. Third, potential data leakage: Stage 1 pseudo-label clustering is run on the same datasets used for final evaluation; the authors never state that test images were excluded from clustering. That needs an explicit split. Fourth, the human evaluation has three radiologists and 50 samples with no agreement or significance statistic, so 'significantly better' is not supported. These are fixable, but together they make the current submission a high-level outline, not a verifiable paper.

Who this is for: researchers working on medical vision-language adaptation might get a useful loss-decomposition idea, and the paper is a good case study for reproducibility standards. I would not cite the numbers.

Recommendation: I would send it to peer review with a clear request for major revision rather than desk-reject. The method is plausible, the equations are well-formed, and the missing items are a concrete checklist: include [5], report uncertainty, name the segmentation model and clustering, release code, and state the data split. If those come back, it could be a decent paper.

Referee Report

4 major / 4 minor

Summary. The paper proposes HiCA (Adaptive Vision-Language Fine-tuning with Hierarchical Contrastive Alignment), a two-stage fine-tuning method for few-shot medical image classification. Stage 1 adapts a vision-language model's visual and text encoders with pseudo-label clustering and a medical text corpus; Stage 2 applies a hierarchical contrastive loss combining global image-text alignment, local ROI-text alignment, and cross-category separation. The authors report experiments on Chest X-ray and Breast Ultrasound datasets, claim state-of-the-art accuracy and AUC in few-shot and zero-shot settings, and include ablation, human evaluation, generalization, robustness, and efficiency analyses.

Significance. If the empirical claims were substantiated, HiCA would be a useful and well-motivated recipe for adapting large vision-language models to medical imaging: the three-level loss decomposition is conceptually clean, the method is described with explicit equations, and the paper addresses a practically important problem. The manuscript also provides a human evaluation, robustness testing, and an efficiency comparison, which are commendable. However, as presented the central claim that HiCA 'consistently outperforms state-of-the-art baselines' is not supported by the evidence: the key state-of-the-art comparator cited in the introduction is absent, no uncertainty quantification or significance statistics appear despite an explicit statement that they were conducted, and several components of the method are underspecified to the point of irreproducibility. The framework may well be defensible, but the paper currently falls short of the evidence bar for its main claim.

major comments (4)

[Section IV-A and Table I] Section IV-A states that experiments cover few-shot settings 'with labeled data ranging from 1-shot to 20-shot per class' and that 'statistical significance testing is conducted to verify the observed improvements,' yet Table I reports only a single point estimate per method per dataset, with no per-shot breakdown, no standard deviations, no confidence intervals, and no p-values. Without variance or repeated-seed information, the claim that HiCA 'consistently outperforms' baselines is an assertion rather than a demonstrated result. This is the central load-bearing evidence for the paper's main conclusion.
[Section I and Section IV-A/Table I] The introduction specifically cites Byra et al. [5] as the prior work that HiCA surpasses (e.g., 'surpassing existing methods by a substantial margin [5]'), but [5] does not appear among the four compared methods listed in Section IV-A nor in Table I. A state-of-the-art claim cannot be evaluated if the stated state-of-the-art comparator is missing from the empirical comparison. The authors need to include [5] (and ideally other recent medical VLM few-shot methods) with the same experimental protocol.
[Section III-B2 and Section III-C] The local alignment loss in Eq. (4) depends on ROIs 'extracted using a segmentation model,' but the segmentation model is never named, and its training data or validation are not described. Likewise, Stage 1 uses 'pseudo-labels generated via unsupervised clustering' and a 'domain-specific corpus of medical texts,' but the clustering algorithm, number of clusters, text corpus, and the base LVLM backbone are all unspecified. These details are required for reproducibility and to rule out that the reported local-alignment gains stem from an unstated segmentation oracle or from overlap between the pseudo-labeling data and the evaluation sets.
[Section IV-B, IV-C, and Table II] The ablation study in Table II is reported only on the Chest X-ray dataset, not on Breast Ultrasound, and the entries are single point estimates without error bars or significance tests. The conclusion in Section IV-C that 'all components are essential for achieving the best performance' is therefore not supported across datasets and could reflect split or seed noise. The authors should provide ablations on both datasets with multiple runs and uncertainty quantification.

minor comments (4)

[Abstract/Conclusion] The abstract and conclusion claim strong performance in 'zero-shot scenarios,' but the experimental section describes only few-shot settings; Table I includes a CLIP zero-shot baseline, but no HiCA zero-shot result is reported. The claim should be either supported with HiCA zero-shot experiments or removed.
[Section IV-D and Table III] The human evaluation is described as involving three radiologists and 50 samples per dataset, but the paper does not report inter-rater agreement, the selection criteria for the 50 samples, whether the radiologists were blinded to method, or any variance in the scores. These details are necessary to interpret the reported interpretability and clinical-validity scores.
[Section IV-E1 and Table IV] The 'unseen categories' generalization experiment is unclear for the two binary datasets used (pneumonia vs. normal and benign vs. malignant); with only two classes, splitting into seen and unseen classes leaves only one class in each set. The setup needs clarification, e.g., whether unseen categories are different pathologies from a larger pool or whether a different dataset construction is used.
[Eq. (4)] The local alignment loss indexes the text embedding as z_text^k, but the notation for text descriptors and class indices is inconsistent with the global loss (where t_i is paired with image x_i). Please clarify the indexing so that the positive and negative pairs in the local loss are unambiguous.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: HiCA is an empirically evaluated method; its losses are defined independently of the reported results, and no prediction reduces to a fitted input or self-citation.

full rationale

The paper's central claim is empirical, not derived: HiCA defines objective terms in Eqs. (2)-(5) and then reports classification accuracy and AUC on standard benchmarks. No equation is fitted to the evaluation numbers, and no baseline result is used as an input to the method. The ablation study in Table II is an empirical check, not a circular proof. The main concerns—omission of the cited SOTA method [5] from Table I, promised significance tests that are never reported, the unnamed ROI segmentation model, and unsupervised pseudo-label clustering on target datasets—are reproducibility or experimental-validity issues rather than circularity: none of them makes the reported state-of-the-art claim true by construction or by definition. There is no load-bearing self-citation (the Y. Zhou references are not by the present authors), no imported uniqueness theorem, and no ansatz smuggled in via citation. The pseudo-label Stage 1 could in principle leak target-distribution information, but the paper does not exhibit a reduction from its equations to its outputs, so under the standard for this analysis no circular step is established.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The paper introduces no new entities. It does introduce several free hyperparameters (lambda1, lambda2, tau, delta, K, number of clusters) whose values are not reported. The key unstated dependency is the segmentation model used for local alignment, which is an ad hoc assumption that the paper does not justify.

free parameters (6)

lambda1 (weight of local alignment loss) = not reported
Introduced in Eq. (2) to balance Llocal; no value or tuning procedure is given in the paper.
lambda2 (weight of cross-category separation loss) = not reported
Introduced in Eq. (2); no value reported.
tau (temperature in contrastive losses) = not reported
Appears in Eqs. (3) and (4) as a temperature scaling cosine similarities; no value given.
delta (margin in cross-category separation) = not reported
Introduced in Eq. (5) as the margin for mismatched pairs; no value reported.
K (number of ROIs per image) = not reported
In Eq. (4), K denotes the number of regions of interest; the paper does not specify K or how ROIs are sampled.
Number of clusters in pseudo-label pretraining = not reported
Stage 1 uses unsupervised clustering to generate pseudo-labels; the clustering algorithm and number of clusters are not given.

assumptions (5)

standard math Contrastive loss backpropagation is well-defined and neural network optimization works as expected.
Assumed by all deep learning methods; not specific to this paper.
domain assumption A pretrained LVLM such as CLIP provides a shared embedding space that can be meaningfully adapted to medical images and text.
The entire method relies on the existence and adaptability of such a model, stated in Section III-A.
domain assumption Pseudo-labels from unsupervised clustering on medical images are informative enough for domain-specific pretraining.
Stage 1 (Section III-C) states the visual encoder is fine-tuned on pseudo-labeled data, but no evidence is provided that the clustering yields meaningful labels.
ad hoc to paper An accurate segmentation model is available to extract ROIs from medical images.
Invoked in Section III-B2 for Llocal; the model is never described, so the accuracy and source of ROIs are unstated assumptions.
domain assumption The two chosen datasets, Chest X-ray and Breast Ultrasound, are representative enough to support general claims of state-of-the-art few-shot performance.
The experiments are limited to two datasets; the paper generalizes beyond this without additional evidence.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Efficient Few-Shot Medical Image Analysis via Hierarchical Contrastive Vision-Language Learning." pith.science (2026). https://pith.science/paper/L76RCWVO

@misc{pith2026250109294,
  author       = {Pith},
  title        = {Pith review of: Efficient Few-Shot Medical Image Analysis via Hierarchical Contrastive Vision-Language Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/L76RCWVO}},
  note         = {Machine review of arXiv:2501.09294}
}

read the original abstract

Few-shot learning in medical image classification presents a significant challenge due to the limited availability of annotated data and the complex nature of medical imagery. In this work, we propose Adaptive Vision-Language Fine-tuning with Hierarchical Contrastive Alignment (HiCA), a novel framework that leverages the capabilities of Large Vision-Language Models (LVLMs) for medical image analysis. HiCA introduces a two-stage fine-tuning strategy, combining domain-specific pretraining and hierarchical contrastive learning to align visual and textual representations at multiple levels. We evaluate our approach on two benchmark datasets, Chest X-ray and Breast Ultrasound, achieving state-of-the-art performance in both few-shot and zero-shot settings. Further analyses demonstrate the robustness, generalizability, and interpretability of our method, with substantial improvements in performance compared to existing baselines. Our work highlights the potential of hierarchical contrastive strategies in adapting LVLMs to the unique challenges of medical imaging tasks.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

22 extracted references · 9 canonical work pages

[5]

Few-shot medical image classification with simple shape and texture text descriptors using vision-language models

M. Byra, M. F. Rachmadi, and H. Skibbe, “Few-shot medical image classiﬁcation with simple shape and texture text descripto rs using vision-language models,” CoRR, vol. abs/2308.04005, 2023. [Online]. Available: https://doi.org/10.48550/arXiv.2308.04005

work page Pith review arXiv doi:10.48550/arxiv.2308.04005 2023
[1]

Learning transferable visual models from na tural language supervision,

A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Aga rwal, G. Sastry, A. Askell, P . Mishkin, J. Clark, G. Krueger, and I. Sutskever, “Learning transferable visual models from na tural language supervision,” in Proceedings of the 38th International Conference on Machine Learning, ICML 2021, 18-24 July 2021, Virtual Event, ser. Proceedings of Machi...

work page 2021
[2]

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

C. Jia, Y . Y ang, Y . Xia, Y . Chen, Z. Parekh, H. Pham, Q. V . Le, Y . Sung, Z. Li, and T. Duerig, “Scaling up visual and vision-language representation learning with noisy text s upervision,” in Proceedings of the 38th International Conference on Machin e Learning, ICML 2021, 18-24 July 2021, Virtual Event , ser. Proceedings of Machine Learning Researc...

work page 2021
[3]

Sketch storytelling,

Y . Zhou, “Sketch storytelling,” in ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICA SSP). IEEE, 2022, pp. 4748–4752

work page 2022
[4]

Style-aware contrastive learning f or multi-style image captioning,

Y . Zhou and G. Long, “Style-aware contrastive learning f or multi-style image captioning,” in Findings of the Association for Computational Linguistics: EACL 2023 , 2023, pp. 2257–2267

work page 2023
[6]

Diffmic: Dual-guidance diffusion network for medical image classiﬁ cation,

Y . Y ang, H. Fu, A. I. Avil´ es-Rivero, C. Sch¨ onlieb, and L. Zhu, “Diffmic: Dual-guidance diffusion network for medical image classiﬁ cation,” in Medical Image Computing and Computer Assisted Interventio n - MICCAI 2023 - 26th International Conference, V ancouver , B C, Canada, October 8-12, 2023, Proceedings, Part VI , ser. Lecture Notes in Computer S...

2023
[7]

Fedmlp: F ederated multi-label medical image classiﬁcation under task hetero geneity,

Z. Sun, N. Wu, J. Shi, L. Y u, K. Cheng, and Z. Y an, “Fedmlp: F ederated multi-label medical image classiﬁcation under task hetero geneity,” in Medical Image Computing and Computer Assisted Interventio n - MICCAI 2024 - 27th International Conference, Marrakesh, Mo rocco, October 6-10, 2024, Proceedings, Part X , ser. Lecture Notes in Computer Science, M....

2024
[8]

Abstract: Interpretable medical image classiﬁcation using prototype learning and privileged inf ormation,

L. Gall´ ee, M. Beer, and M. G¨ otz, “Abstract: Interpretable medical image classiﬁcation using prototype learning and privileged inf ormation,” in Bildverarbeitung f¨ ur die Medizin 2024 - Proceedings, Germ an Conference on Medical Image Computing, Erlangen, March 10- 12, 2024, ser. Informatik Aktuell, A. Maier, T. M. Deserno, H. Handel s, K. H. Maier-He...

work page doi:10.1007/978-3-65 2024

Show all 22 references

[9]

Robu st and interpretable medical image classiﬁers via concept bot tleneck models,

A. Y an, Y . Wang, Y . Zhong, Z. He, P . Karypis, Z. Wang, C. Dong, A. Gentili, C. Hsu, J. Shang, and J. J. McAuley, “Robu st and interpretable medical image classiﬁers via concept bot tleneck models,” CoRR, vol. abs/2310.03182, 2023. [Online]. Available: https://doi.org/10.48...

doi arXiv
[10]

Enhancing multimodal medical image classiﬁcation using cross-graph modal contr astive learning,

J. Ding, C. Hsu, and F. Liu, “Enhancing multimodal medical image classiﬁcation using cross-graph modal contr astive learning,” CoRR, vol. abs/2410.17494, 2024. [Online]. Available: https://doi.org/10.48550/arXiv.2410.17494

2024 doi
[11]

Visual in-context l earning for large vision-language models,

Y . Zhou, X. Li, Q. Wang, and J. Shen, “Visual in-context l earning for large vision-language models,” in Findings of the Association for Computational Linguistics, ACL 2024, Bangkok, Thailand an d virtual meeting, August 11-16, 2024 . Association for Computational Linguis- ti...

2024
[12]

Medmamba: Vision mamba for medical ima ge classiﬁcation,

Y . Y ue and Z. Li, “Medmamba: Vision mamba for medical ima ge classiﬁcation,” CoRR, vol. abs/2403.03849, 2024. [Online]. Available: https://doi.org/10.48550/arXiv.2403.03849

2024 doi arXiv
[13]

Improv ing zero-shot cross-lingual transfer for multilingual questi on answering over knowledge graph,

Y . Zhou, X. Geng, T. Shen, W. Zhang, and D. Jiang, “Improv ing zero-shot cross-lingual transfer for multilingual questi on answering over knowledge graph,” in Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Ling uistics: H...

2021
[14]

Medvilam: A multimodal large language model with advanced generalizability and explainability for medical data unde rstanding and generation,

L. Xu, H. Sun, Z. Ni, H. Li, and S. Zhang, “Medvilam: A multimodal large language model with advanced generalizability and explainability for medical data unde rstanding and generation,” CoRR, vol. abs/2409.19684, 2024. [Online]. Available: https://doi.org/10.48550/arXiv.2409.19684

2024 doi arXiv
[15]

Less is more: Vision representation compression for efﬁcient video gene ration with large language models,

Y . Zhou, J. Zhang, G. Chen, J. Shen, and Y . Cheng, “Less is more: Vision representation compression for efﬁcient video gene ration with large language models,” 2024

2024
[16]

Thread of thought unraveling chaotic contexts,

Y . Zhou, X. Geng, T. Shen, C. Tao, G. Long, J.-G. Lou, and J. Shen, “Thread of thought unraveling chaotic contexts,” arXiv preprint arXiv:2311.08734, 2023

2023 arXiv
[17]

Medica l adaptation of large language and vision-language models: Are we making pr ogress?

D. P . Jeong, S. Garg, Z. C. Lipton, and M. Oberst, “Medica l adaptation of large language and vision-language models: Are we making pr ogress?” in Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, EMNLP 2024, Miami, FL, USA, November 12-1...

2024
[18]

On large visual language models for medical imaging analysis: An empirical study,

M. V an, P . V erma, and X. Wu, “On large visual language models for medical imaging analysis: An empirical study,” i n IEEE/ACM Conference on Connected Health: Applications, Sy stems and Engineering Technologies, CHASE 2024, Wilmington, DE, USA, June 19-21, 2024 . IEEE, 2024,...

2024
[19]

Training medical large vi sion- language models with abnormal-aware feedback,

Y . Zhou, L. Song, and J. Shen, “Training medical large vi sion- language models with abnormal-aware feedback,” arXiv preprint arXiv:2501.01377, 2025

2025 arXiv
[20]

Prompting medical large vis ion- language models to diagnose pathologies by visual question answering,

D. Guo and D. Terzopoulos, “Prompting medical large vis ion- language models to diagnose pathologies by visual question answering,” CoRR, vol. abs/2407.21368, 2024. [Online]. Available: https://doi.org/10.48550/arXiv.2407.21368

2024 doi arXiv
[21]

GMAI-VL & GMAI-VL-5.5M: A large vision-language model and A comprehensive multimodal dataset towards gener al medical AI,

T. Li, Y . Su, W. Li, B. Fu, Z. Chen, Z. Huang, G. Wang, C. Ma, Y . Chen, M. Hu, Y . Li, P . Chen, X. Hu, Z. Deng, Y . Ji, J. Y e, Y . Qia o, and J. He, “GMAI-VL & GMAI-VL-5.5M: A large vision-language model and A comprehensive multimodal dataset towards gener al medical AI,” ...

2024 doi arXiv
[22]

Improving cross-modal alignment f or text- guided image inpainting,

Y . Zhou and G. Long, “Improving cross-modal alignment f or text- guided image inpainting,” in Proceedings of the 17th Conference of the European Chapter of the Association for Computational L inguistics, 2023, pp. 3445–3456. 5

2023

Pith tools

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

[1] [5]

Few-shot medical image classification with simple shape and texture text descriptors using vision-language models

M. Byra, M. F. Rachmadi, and H. Skibbe, “Few-shot medical image classiﬁcation with simple shape and texture text descripto rs using vision-language models,” CoRR, vol. abs/2308.04005, 2023. [Online]. Available: https://doi.org/10.48550/arXiv.2308.04005

work page Pith review arXiv doi:10.48550/arxiv.2308.04005 2023

[2] [1]

Learning transferable visual models from na tural language supervision,

A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Aga rwal, G. Sastry, A. Askell, P . Mishkin, J. Clark, G. Krueger, and I. Sutskever, “Learning transferable visual models from na tural language supervision,” in Proceedings of the 38th International Conference on Machine Learning, ICML 2021, 18-24 July 2021, Virtual Event, ser. Proceedings of Machi...

work page 2021

[3] [2]

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

C. Jia, Y . Y ang, Y . Xia, Y . Chen, Z. Parekh, H. Pham, Q. V . Le, Y . Sung, Z. Li, and T. Duerig, “Scaling up visual and vision-language representation learning with noisy text s upervision,” in Proceedings of the 38th International Conference on Machin e Learning, ICML 2021, 18-24 July 2021, Virtual Event , ser. Proceedings of Machine Learning Researc...

work page 2021

[4] [3]

Sketch storytelling,

Y . Zhou, “Sketch storytelling,” in ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICA SSP). IEEE, 2022, pp. 4748–4752

work page 2022

[5] [4]

Style-aware contrastive learning f or multi-style image captioning,

Y . Zhou and G. Long, “Style-aware contrastive learning f or multi-style image captioning,” in Findings of the Association for Computational Linguistics: EACL 2023 , 2023, pp. 2257–2267

work page 2023

[6] [6]

Diffmic: Dual-guidance diffusion network for medical image classiﬁ cation,

Y . Y ang, H. Fu, A. I. Avil´ es-Rivero, C. Sch¨ onlieb, and L. Zhu, “Diffmic: Dual-guidance diffusion network for medical image classiﬁ cation,” in Medical Image Computing and Computer Assisted Interventio n - MICCAI 2023 - 26th International Conference, V ancouver , B C, Canada, October 8-12, 2023, Proceedings, Part VI , ser. Lecture Notes in Computer S...

2023

[7] [7]

Fedmlp: F ederated multi-label medical image classiﬁcation under task hetero geneity,

Z. Sun, N. Wu, J. Shi, L. Y u, K. Cheng, and Z. Y an, “Fedmlp: F ederated multi-label medical image classiﬁcation under task hetero geneity,” in Medical Image Computing and Computer Assisted Interventio n - MICCAI 2024 - 27th International Conference, Marrakesh, Mo rocco, October 6-10, 2024, Proceedings, Part X , ser. Lecture Notes in Computer Science, M....

2024

[8] [8]

Abstract: Interpretable medical image classiﬁcation using prototype learning and privileged inf ormation,

L. Gall´ ee, M. Beer, and M. G¨ otz, “Abstract: Interpretable medical image classiﬁcation using prototype learning and privileged inf ormation,” in Bildverarbeitung f¨ ur die Medizin 2024 - Proceedings, Germ an Conference on Medical Image Computing, Erlangen, March 10- 12, 2024, ser. Informatik Aktuell, A. Maier, T. M. Deserno, H. Handel s, K. H. Maier-He...

work page doi:10.1007/978-3-65 2024