Pith. sign in

REVIEW 3 major objections 4 minor 38 references

Many-MobileNet: Multi-Model Augmentation for Robust Retinal Disease Classification

T0 review · 3 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read This paper claims that fusing three nnMobileNet models—two lightweight and one medium-sized, each trained with a different data augmentation strategy—yields better generalization in data-scarce retinal image quality classification, as…

desk verdict The paper's own validation table contradicts its fusion-benefit claim, and with no test-set baselines for the single models, the rank-3 result cannot be attributed to fusion. read the letter →

arxiv 2412.02825 v1 pith:VD2YGP5Y submitted 2024-12-03 cs.CV

Hao Wang , Wenhui Zhu , Xuanzhao Dong , Yanxi Chen , Xin Li , Peijie Qiu , Xiwen Chen , Vamsi Krishna Vasa
show 4 more authors
Yujian Xiong Oana M. Dumitrascu Abolfazl Razi Yalin Wang
This is my paper · ORCID
classification cs.CV
keywords retinaldiseaseclassificationfundusimagequalityassessmentultra-widefieldimagingdiabeticretinopathymodelfusionnnMobileNetdataaugmentationlightweightCNN
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 tries to establish that a small ensemble of nnMobileNet models can generalize better than any single model when classifying the quality of ultra-widefield fundus images from a dataset of only 434 samples. The authors train two lightweight versions with channel multiplier 1.0 and one medium-sized version with multiplier 3.0, give each a distinct data augmentation strategy, and fuse their predictions by max or average voting. They enter the fused models in the UWF4DR image quality assessment task and report a third-place official test result, with AUROC 0.9525 and AUPRC 0.9683. The practical payoff would be a lightweight, compute-cheap recipe for medical imaging tasks where data is scarce and a single network tends to overfit.

What carries the argument

The mechanism is Many-MobileNet, a fusion of three instances of the nnMobileNet CNN, a lightweight architecture using depthwise separable convolutions, linear bottlenecks, and squeeze-and-excitation attention blocks. The three instances differ in channel width (two at width 1.0, one at width 3.0) and in data augmentation, mainly through different normalization values; during inference their outputs are combined by max or average voting. Model width is what varies capacity: width 1.0 gives a small model of roughly 13 MB, while width 3.0 triples the channels in each layer to produce a model of roughly 120 MB, so the fusion pairs speed with higher-capacity feature extraction. This combination of architectural diversity and augmentation diversity is what the paper claims carries the generalization benefit.

What would settle it

A reader could test the claim by scoring each of the three constituent models separately on the official test set; if any single model reaches or exceeds the fused AUROC of 0.9525, the generalization benefit attributed to fusion is not demonstrated.

Watch

Extended reading notes

Core claim

The central claim is that fusing multiple nnMobileNet classifiers trained on the same small dataset with deliberately different augmentation schemes and model widths reduces overfitting and improves test-time generalization. The authors report that on the official UWF4DR validation set, the ensemble actually scored lower than its best individual members, and they attribute this to the validation set being small and potentially unrepresentative. On the official test set, the ensemble placed third, with balanced sensitivity and specificity, and the authors take this as evidence that the fusion generalizes better than any single configuration. In their telling, the diversity among members—two efficient width-1.0 models and one width-3.0 model with separate normalization values—lets each model learn complementary aspects of the fundus images, so voting combines their strengths.

Load-bearing premise

The argument rests on believing that the ensemble's worse score on the validation set is due to the validation set being small and unrepresentative, and that the third-place official test result is real evidence of fusion's benefit rather than a chance property of that particular test split.

Editorial extensions

If this is right

  • If the central claim is correct, the same multi-width, multi-augmentation fusion recipe can be transferred to other small medical image datasets where a single network overfits.
  • The paper's reported inference time of 0.1098 seconds per image shows the fused system remains practical for screening despite running three models.
  • The official test AUROC of 0.9525, paired with the weaker validation result, implies that validation-based model selection would have discarded the configuration that performed best on the test set.
  • The fusion design means a deployed system can include a fast width-1.0 model and still keep the higher-capacity width-3.0 model in the final vote.

Reading between the lines

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

  • Editorial inference: the pattern of results suggests that on UWF4DR-style small validation splits, model selection by validation score can systematically disfavor ensembles that average out member errors; this could be tested by re-running selection on multiple random validation folds.
  • Editorial inference: since only two widths and three models were used, the fusion gain could be probed further by varying the number of members, the width ratio, and the augmentation policies, with a fixed compute budget, to see where the benefit saturates.
  • Editorial inference: the stated reason for member diversity is different normalization values; a clean experiment would hold all else equal, swap only the normalization statistics between members, and measure whether the fusion gain survives.
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

3 major / 4 minor

Summary. The paper describes Many-MobileNet, an ensemble of three nnMobileNet classifiers (two width-1.0 models and one width-3.0 model) trained on the UWF4DR image-quality task with different data augmentation, normalization, and hyperparameter settings, and combined at inference by max or average voting. The authors report that this fusion secured third place on the official UWF4DR test set (AUROC 0.9525) and claim in the abstract and conclusion that fusing multiple models improves robustness and generalization in data-scarce retinal image classification. The manuscript includes validation results for the three component models and the ensemble, plus the official challenge ranking for the fused submission.

Significance. If the central claim were established, the paper would provide a simple and computationally efficient ensemble recipe for small medical-image datasets, with the external validation of a third-place challenge ranking and public code as supporting assets. However, the key benefit of fusion over single models is not demonstrated: the only internal comparison (Table 2) shows the ensemble underperforming every component on validation AUROC/AUPRC, and no test-set results are reported for individual models. As a result, the paper currently illustrates a selection/aggregation pitfall rather than a validated method, so its contribution is not yet established.

major comments (3)
  1. [Section 3.2, Table 2] Table 2 contradicts the central claim. On the official validation set, Ensemble with max has AUROC 0.8468 and Ensemble with average has AUROC 0.8125, both below every single model (Weight1 0.8772, Weight2 0.8626, Weight3 0.8581); the same ordering holds for AUPRC. The abstract and Section 4 state that fusion 'significantly improved the model's generalization,' but the only within-paper comparison shows fusion degrading the primary metric. The explanation in Section 3.2 that the validation set is 'small and potentially unrepresentative' is an assertion, not evidence, and it is in tension with the paper's own use of that validation set to select the three component models.
  2. [Section 3.2, Table 3] No test-set baselines for the component models are reported. Table 3 contains only the fused submission (AUROC 0.9525, rank 3). Without the AUROC/AUPRC of Weight1, Weight2, and Weight3 on the official test set, or a non-fused baseline, the rank-3 result cannot be attributed to fusion. It could equally reflect the strength of one of the selected models, model-selection effects on a small validation set, or test-set distribution differences. A counterfactual comparison (e.g., each single model scored on the test set, or repeated cross-validation with and without fusion) is required to support the paper's conclusion.
  3. [Section 2.5 and Section 3.2] The fusion rule is under-specified. Section 2.5 says the final decision is 'the maximum or average outputs' of the three models, and Table 2 reports results for both, but the submitted system's choice of rule is not stated. Moreover, the paper does not specify how the models' outputs are calibrated before averaging or maxing; if raw logits or uncalibrated probabilities are combined, the max rule is sensitive to scale differences between the width-1.0 and width-3.0 models. This matters because Table 2 shows the two rules differ by 0.034 AUROC, and the chosen rule is part of the method being evaluated.
minor comments (4)
  1. [Section 2.5] The text says '13M Bin' and '120M B' where 'MB' is intended; these typos should be corrected.
  2. [Section 2.3 and Table 1] The text states dropout rates from 0 to 0.10 were tested, but Table 1 lists only 0.01, 0.02, and 0.05; the discrepancy should be reconciled.
  3. [Figure 4] Figure 4 is described as 'Parameter empirical studies' but no legend or detailed axis labels are provided, so the reader cannot determine which curves correspond to which hyperparameter.
  4. [Section 2.1] The name 'Many-MobileNet' is used for an ensemble of existing nnMobileNet models; for clarity, the title and text should distinguish the architecture from the fusion strategy.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the rank-3 result is an external benchmark, and the fusion-improvement claim is an evidence gap rather than a self-referential derivation.

full rationale

The paper makes no first-principles derivation and contains no fitted constant that is later relabeled as a prediction. Its central evidence is an external benchmark: the UWF4DR official test-set score (AUROC 0.9525, rank 3), which is computed by the challenge organizers and is not a function of the paper's own definitions or validation claims. The only internal comparison, Table 2, shows the fused models underperforming each component on validation, and Section 3.2 explains this with an empirical guess ('This performance decrement might be caused by the small and potentially unrepresentative validation data') before concluding that fusion improved generalization; that is an unsupported inference and a missing ablation, but it is not circular, because the official test result does not reduce to the validation explanation or to the fusion formula. Self-citations (nnMobileNet, prior retinal work) supply the fixed architecture, not the target claim, and no uniqueness theorem or hidden ansatz is imported to force the result. Therefore no circular step is identifiable; the correctness risk is about missing component-level test baselines, not circularity.

Assumptions & free parameters 7 free parameters · 4 assumptions · 0 invented entities

The method rests on hyperparameters selected by validation performance, an underspecified augmentation and normalization scheme, and the assumption that validation and test outcomes are comparable. No new entities are introduced.

free parameters (7)
  • learning rate per model = 1e-5, 1e-5, 1e-4
    Chosen by validation-set performance; listed in Table 2.
  • batch size per model = 8, 16, 16
    Chosen by validation-set performance; listed in Table 2.
  • dropout rate per model = 0.01, 0.02, 0.05
    Chosen by validation-set performance; listed in Table 2.
  • model width (channel multiplier) = 1.0, 1.0, 3.0
    Chosen to balance efficiency and capacity; listed in Table 2.
  • data augmentation/normalization scheme per model = not specified in paper
    The paper states different normalization strategies are used but does not specify them; these choices are fitted to the task.
  • weight decay = 0.005
    Fixed in Table 1; not clear if tuned.
  • fusion rule = max or average
    Both are evaluated; neither is justified by validation performance.
assumptions (4)
  • domain assumption The 434-sample validation split is representative enough for model selection and generalization claims.
    The paper relies on this split to select models; the authors later call it potentially unrepresentative (Section 3.2).
  • domain assumption The official test set ranking is produced by the described fusion of the top three validation models.
    No test-set results for individual models are reported, so this assumption is needed to attribute the score to fusion.
  • domain assumption nnMobileNet is an appropriate base architecture for this task.
    Taken from the authors' prior work [6]; no independent architectural comparison in this paper.
  • standard math Cross-entropy loss and AdamP with cosine decay are suitable training choices.
    Standard supervised learning tools; not central to the claim.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Many-MobileNet: Multi-Model Augmentation for Robust Retinal Disease Classification." pith.science (2026). https://pith.science/paper/VD2YGP5Y

@misc{pith2026241202825,
  author       = {Pith},
  title        = {Pith review of: Many-MobileNet: Multi-Model Augmentation for Robust Retinal Disease Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VD2YGP5Y}},
  note         = {Machine review of arXiv:2412.02825}
}
read the original abstract

In this work, we propose Many-MobileNet, an efficient model fusion strategy for retinal disease classification using lightweight CNN architecture. Our method addresses key challenges such as overfitting and limited dataset variability by training multiple models with distinct data augmentation strategies and different model complexities. Through this fusion technique, we achieved robust generalization in data-scarce domains while balancing computational efficiency with feature extraction capabilities.

Figures

Figures reproduced from arXiv: 2412.02825 by the authors.

Figure 1
Figure 1. Concept of model fusion during inference. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Prediction of Many-MobileNet. size of around 120MB. This makes the model width a key factor in balancing between computational efficiency and the ability to capture complex features. Our model fusion involves the combination of multiple nnMobileNet models, each deployed with different channel multipliers. We employ two lightweight models with a channel multiplier of 1, optimized for efficiency, and one medium-sized … view at source ↗
Figure 3
Figure 3. Sample images of UWF4DR dataset, where label 0 represents the ungradable images and label 1 represents [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Parameter empirical studies based on UWF4DR dataset. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 32 canonical work pages

  1. [1]

    Retinal diseases and vision 2020

    David Yorston. Retinal diseases and vision 2020. Community Eye Health, 16(46):19–20, 2003

    work page 2020

  2. [2]

    Beyond mobilenet: An improved mobilenet for retinal diseases

    Wenhui Zhu, Peijie Qiu, Xiwen Chen, Huayu Li, Hao Wang, Natasha Lepore, Oana M Dumitrascu, and Yalin Wang. Beyond mobilenet: An improved mobilenet for retinal diseases. In International Conference on Medical Image Computing and Computer-Assisted Intervention, pages 56–65. Springer, 2023

    work page 2023

  3. [3]

    Global prevalence of myopia and high myopia and temporal trends from 2000 through 2050

    Brien A Holden, Timothy R Fricke, David A Wilson, Monica Jong, Kovin S Naidoo, Padmaja Sankaridurg, Tien Y Wong, Thomas J Naduvilath, and Serge Resnikoff. Global prevalence of myopia and high myopia and temporal trends from 2000 through 2050. Ophthalmology, 123(5):1036–1042, 2016. 6 H. Wang et al

    work page 2000

  4. [4]

    Rbad: A dataset and benchmark for retinal vessels branching angle detection

    Hao Wang, Wenhui Zhu, Jiayou Qin, Xin Li, Oana Dumitrascu, Xiwen Chen, Peijie Qiu, and Abolfazl Razi. Rbad: A dataset and benchmark for retinal vessels branching angle detection. arXiv preprint arXiv:2407.12271, 2024

    arXiv 2024

  5. [5]

    Diabetic retinopathy detection through integration of deep learning classification framework

    Alexander Rakhlin. Diabetic retinopathy detection through integration of deep learning classification framework. BioRxiv, page 225508, 2017

    work page 2017

  6. [6]

    Dumitrascu, and Yalin Wang

    Wenhui Zhu, Peijie Qiu, Xiwen Chen, Xin Li, Natasha Lepore, Oana M. Dumitrascu, and Yalin Wang. nnmobilenet: Rethinking cnn for retinopathy research. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, pages 2285–2294, June 2024

    work page 2024

  7. [7]

    V o and Abhishek Verma

    Holly H. V o and Abhishek Verma. New deep neural nets for fine-grained diabetic retinopathy recognition on hybrid color space. In 2016 IEEE International Symposium on Multimedia (ISM), pages 209–215, 2016

    work page 2016

  8. [8]

    Image processing and classification in diabetic retinopathy: A review

    Arslan Ahmad, Atif Bin Mansoor, Rafia Mumtaz, Mukaram Khan, and SH Mirza. Image processing and classification in diabetic retinopathy: A review. In 2014 5th European workshop on visual information processing (EUVIP), pages 1–6. IEEE, 2014

    work page 2014

Show all 38 references
  1. [9]

    Fast key points detection and matching for tree- structured images

    Hao Wang, Xiwen Chen, Abolfazl Razi, and Rahul Amin. Fast key points detection and matching for tree- structured images. In 2022 International Conference on Computational Science and Computational Intelligence (CSCI), pages 1381–1387. IEEE, 2022

    2022

  2. [10]

    Green: a graph residual re-ranking network for grading diabetic retinopathy

    Shaoteng Liu, Lijun Gong, Kai Ma, and Yefeng Zheng. Green: a graph residual re-ranking network for grading diabetic retinopathy. In MICCAI, pages 585–594, Cham, 2020. Springer International Publishing

    2020

  3. [11]

    Evaluation of a computer-aided diagnosis system for diabetic retinopathy screening on public data

    Clara I Sánchez and et al. Evaluation of a computer-aided diagnosis system for diabetic retinopathy screening on public data. Investigative ophthalmology & visual science, 52(7):4866–4871, 2011

    2011

  4. [12]

    Automated retinal imaging analysis for alzheimers disease screening

    Oana M Dumitrascu, Wenhui Zhu, Peijie Qiu, Keshav Nandakumar, and Yalin Wang. Automated retinal imaging analysis for alzheimers disease screening. In IEEE International Symposium on Biomedical Imaging: From Nano to Macro (ISBI), 2022

    2022

  5. [13]

    Mayo Clinic Proceedings: Digital Health, 2024

    Color fundus photography and deep learning applications in alzheimer’s disease. Mayo Clinic Proceedings: Digital Health, 2024

    2024

  6. [14]

    Robust pca with lw, and l2, 1 norms: A novel method for low-quality retinal image enhancement

    Habte Tadesse Likassa, Ding-Geng Chen, Kewei Chen, Yalin Wang, and Wenhui Zhu. Robust pca with lw, and l2, 1 norms: A novel method for low-quality retinal image enhancement. Journal of Imaging, 10(7):151, 2024

    2024

  7. [15]

    A deep learning system for detecting diabetic retinopathy across the disease spectrum

    Ling Dai, Liang Wu, Huating Li, Chun Cai, Qiang Wu, Hongyu Kong, Ruhan Liu, Xiangning Wang, Xuhong Hou, Yuexing Liu, et al. A deep learning system for detecting diabetic retinopathy across the disease spectrum. Nature communications, 12(1):3242, 2021

    2021

  8. [16]

    Deepdrid: Diabetic retinopathy—grading and image quality estimation challenge

    Ruhan Liu, Xiangning Wang, Qiang Wu, Ling Dai, Xi Fang, Tao Yan, Jaemin Son, Shiqi Tang, Jiang Li, Zijian Gao, et al. Deepdrid: Diabetic retinopathy—grading and image quality estimation challenge. Patterns, 3(6), 2022

    2022

  9. [17]

    Context-aware optimal transport learning for retinal fundus image enhancement

    Vamsi Krishna Vasa, Peijie Qiu, Wenhui Zhu, Yujian Xiong, Oana Dumitrascu, and Yalin Wang. Context-aware optimal transport learning for retinal fundus image enhancement. arXiv preprint arXiv:2409.07862, 2024

    2024 arXiv

  10. [18]

    X. Li, X. Hu, L. Yu, L. Zhu, C. W. Fu, and P. A. Heng. CANet: Cross-Disease Attention Network for Joint Diabetic Retinopathy and Diabetic Macular Edema Grading. IEEE Trans Med Imaging, pages 1483–1493, 2020

    2020

  11. [19]

    Zoom-in-net: Deep mining lesions for diabetic retinopathy detection

    Zhe Wang, Yanxin Yin, Jianping Shi, Wei Fang, Hongsheng Li, and Xiaogang Wang. Zoom-in-net: Deep mining lesions for diabetic retinopathy detection. In MICCAI, pages 267–275, 2017

    2017

  12. [20]

    Collaborative learning of semi-supervised segmentation and classification for medical images

    Yi Zhou, Xiaodong He, Lei Huang, Li Liu, Fan Zhu, Shanshan Cui, and Ling Shao. Collaborative learning of semi-supervised segmentation and classification for medical images. In Proc. IEEE Comput. Soc. Conf. Comput. Vis. Pat- tern Recognit, 2019

    2019

  13. [21]

    Chen, and Jian Wu

    Zhiwen Lin, Ruoqian Guo, Yanjie Wang, Bian Wu, Tingting Chen, Wenzhe Wang, Danny Z. Chen, and Jian Wu. A framework for identifying diabetic retinopathy based on anti-noise detection and attention-based fusion. In MICCAI, pages 74–82, Cham, 2018. Springer

    2018

  14. [22]

    Learning robust representation for joint grading of ophthalmic diseases via adaptive curriculum and feature disentanglement

    Haoxuan Che, Haibo Jin, and Hao Chen. Learning robust representation for joint grading of ophthalmic diseases via adaptive curriculum and feature disentanglement. In MICCAI, pages 523–533, 2022

    2022

  15. [23]

    Self-supervised equivariant regularization reconciles multiple instance learning: Joint referable diabetic retinopathy classification and lesion segmentation

    Wenhui Zhu and et al. Self-supervised equivariant regularization reconciles multiple instance learning: Joint referable diabetic retinopathy classification and lesion segmentation. 18th International Symposium on Medical Information Processing and Analysis (SIPAIM), 2022

    2022

  16. [24]

    Transformers in vision: A survey

    Salman Khan, Muzammal Naseer, Munawar Hayat, Syed Waqas Zamir, Fahad Shahbaz Khan, and Mubarak Shah. Transformers in vision: A survey. ACM computing surveys (CSUR), 54(10s):1–41, 2022. 7 H. Wang et al

    2022

  17. [25]

    Otre: Where optimal transport guided unpaired image-to-image translation meets regularization by enhancing

    Wenhui Zhu, Peijie Qiu, Oana M Dumitrascu, Jacob M Sobczak, Mohammad Farazi, Zhangsihao Yang, Keshav Nandakumar, and Yalin Wang. Otre: Where optimal transport guided unpaired image-to-image translation meets regularization by enhancing. In International Conference on Informati...

    2023

  18. [26]

    Optimal transport guided unsupervised learning for enhancing low-quality retinal images.arXiv preprint arXiv:2302.02991, 2023

    Wenhui Zhu, Peijie Qiu, Mohammad Farazi, Keshav Nandakumar, Oana M Dumitrascu, and Yalin Wang. Optimal transport guided unsupervised learning for enhancing low-quality retinal images.arXiv preprint arXiv:2302.02991, 2023

    2023 arXiv

  19. [27]

    Exploring the limits of data augmentation for retinal vessel segmentation

    Enes Sadi Uysal, M ¸ Safak Bilici, B Selin Zaza, M Yi˘git Özgenç, and Onur Boyar. Exploring the limits of data augmentation for retinal vessel segmentation. arXiv preprint arXiv:2105.09365, 2021

    2021 arXiv

  20. [28]

    Using mri-specific data augmentation to enhance the segmentation of right ventricle in multi-disease, multi-center and multi-view cardiac mri

    Tewodros Weldebirhan Arega, François Legrand, Stéphanie Bricq, and Fabrice Meriaudeau. Using mri-specific data augmentation to enhance the segmentation of right ventricle in multi-disease, multi-center and multi-view cardiac mri. In International Workshop on Statistical Atlase...

    2021

  21. [29]

    Mobilenetv2: Inverted residuals and linear bottlenecks

    Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zhmoginov, and Liang-Chieh Chen. Mobilenetv2: Inverted residuals and linear bottlenecks. In Proc. IEEE Comput. Soc. Conf. Comput. Vis. Pat- tern Recognit, pages 4510–4520, 2018

    2018

  22. [30]

    Efficient object localization using convolutional networks

    Jonathan Tompson, Ross Goroshin, Arjun Jain, Yann LeCun, and Christoph Bregler. Efficient object localization using convolutional networks. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 648–656, 2015

    2015

  23. [31]

    Cbam: Convolutional block attention module

    Sanghyun Woo, Jongchan Park, Joon-Young Lee, and In So Kweon. Cbam: Convolutional block attention module. In Proceedings of the European conference on computer vision (ECCV), pages 3–19, 2018

    2018

  24. [32]

    Rethinking channel dimensions for efficient model design

    Dongyoon Han, Sangdoo Yun, Byeongho Heo, and YoungJoon Yoo. Rethinking channel dimensions for efficient model design. In Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition, pages 732–741, 2021

    2021

  25. [33]

    Squeeze-and-excitation networks

    Jie Hu, Li Shen, and Gang Sun. Squeeze-and-excitation networks. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 7132–7141, 2018

    2018

  26. [34]

    Lesion-aware transformers for diabetic retinopathy grading

    Rui Sun, Yihao Li, Tianzhu Zhang, Zhendong Mao, Feng Wu, and Yongdong Zhang. Lesion-aware transformers for diabetic retinopathy grading. In Proc. IEEE Comput. Soc. Conf. Comput. Vis. Pat- tern Recognit , pages 10938–10947, 2021

    2021

  27. [35]

    Satformer: Saliency-guided abnormality-aware transformer for retinal disease classification in fundus image

    Yankai Jiang and et al. Satformer: Saliency-guided abnormality-aware transformer for retinal disease classification in fundus image. In Proceedings of the Thirty-First International Joint Conference on Artificial Intelligence, IJCAI, pages 987–994, 2022

    2022

  28. [36]

    A convnet for the 2020s

    Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, and Saining Xie. A convnet for the 2020s. In Proc. IEEE Comput. Soc. Conf. Comput. Vis. Pat- tern Recognit, pages 11976–11986, 2022

    2022

  29. [37]

    Mil-vt: Multiple instance learning enhanced vision transformer for fundus image classification

    Shuang Yu and et al. Mil-vt: Multiple instance learning enhanced vision transformer for fundus image classification. In MICCAI, pages 45–54. Springer, 2021

    2021

  30. [38]

    Adamp: Slowing down the slowdown for momentum optimizers on scale-invariant weights

    Byeongho Heo, Sanghyuk Chun, Seong Joon Oh, Dongyoon Han, Sangdoo Yun, Gyuwan Kim, Youngjung Uh, and Jung-Woo Ha. Adamp: Slowing down the slowdown for momentum optimizers on scale-invariant weights. arXiv preprint arXiv:2006.08217, 2020. 8

    2006 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.