Pith. sign in

REVIEW 4 major objections 5 minor 46 references

One-Shot Neural Architecture Search with Network Similarity Directed Initialization for Pathological Image Classification

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Domain adaptation added to one-shot neural architecture search makes validation scores predict test performance on breast pathology images.

desk verdict Solid idea, undercut by epoch mismatch: the MMD alignment idea is worth testing, but the F1 claims are confounded. read the letter →

arxiv 2506.14176 v1 pith:ZSOBCDUM submitted 2025-06-17 cs.CV

Renao Yan This is my paper · ORCID
classification cs.CV
keywords neuralarchitecturesearchone-shotNASdomainadaptationmaximummeandiscrepancycomputationalpathologybreastcancerclassificationevolutionaryalgorithmpopulationdiversity
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 argues that one-shot neural architecture search (NAS) for pathological image classification is held back by two problems: evolutionary search starts from poorly diversified populations, and supernet rankings from training-set features do not transfer to the test distribution. It proposes DAOS, which adds a Network Similarity Directed Initialization (NSDI) that forces the initial architectures to be diverse, and a Maximum Mean Discrepancy (MMD) loss during supernet training that aligns labeled training features with unlabeled validation features. On the BRACS breast histology dataset, the method reports a best F1 of 61.41% versus 59.54% for the SPOS baseline, and raises the Pearson correlation between validation and test F1 from 0.1794 to roughly 0.70. If correct, this means that stable ranking in medical-image NAS does not require extra search cost, and that domain adaptation can be grafted onto existing one-shot pipelines.

What carries the argument

The load-bearing machinery is a pair of mechanisms in the DAOS pipeline. First, Network Similarity Directed Initialization (NSDI) encodes each candidate architecture as a binary vector, defines pairwise similarity $SS(v_i,v_j)=\sum_{k=1}^{N} v_{i,k}\odot v_{j,k}$ with $\odot$ the XNOR operation, and constrains the Average Population Similarity to stay below a threshold $APS_{\max}$ while sampling, with a timeout that relaxes the threshold if sampling stalls. This forces the initial population to cover the discrete search space instead of clustering. Second, the supernet training objective augments the classification cross-entropy with an MMD term $\lambda\sum_{\gamma\in\{q,l\}}\hat{d}_{\mathcal{H}}(p,\gamma)$ over unlabeled validation features, using the unbiased kernel estimator of Equation (9). The MMD term is what aligns training and validation distributions, and it is this alignment that the paper credits for the improved validation–test correlation.

What would settle it

Retrain the DAOS-A supernet with the MMD loss but evaluate validation-to-test correlation on a held-out split that was never used as MMD target or for architecture selection; if the Pearson correlation drops back toward the 0.18 baseline, the central claim of improved ranking reliability fails.

Watch

Extended reading notes

Core claim

The central claim is that injecting domain adaptation into one-shot NAS jointly fixes search stability and evaluation reliability on histopathology images. With NSDI, the initial population is chosen so that no architecture is too similar to another, measured by an XNOR overlap of their operation encodings, avoiding the unbalanced exploration that random initialization produces under small population budgets. With the MMD loss, the supernet is trained to produce features that are indistinguishable between the training and validation splits, so that validation F1 becomes a trustworthy predictor of test F1. The paper reports that DAOS-A achieves 61.41% F1 on BRACS, exceeding SPOS (59.54%) and the random-search baseline, and that the validation–test Pearson correlation rises from 0.1794 to 0.6985 (DAOS-A) and 0.7096 (DAOS-B). It further claims this is the first incorporation of domain adaptation into one-shot NAS and the first explicit quantification of population diversity during initialization.

Load-bearing premise

The load-bearing premise is that the BRACS training and validation splits really do differ in staining and scale in a way that MMD alignment corrects, and that using the validation split both as the unlabeled adaptation target and as the selector of architectures does not make the reported correlation improvement a measure of the method's ability to predict unseen test performance independently.

Editorial extensions

If this is right

  • If the ranking improvement is real, one-shot NAS becomes usable for histopathology without paying extra retraining cost, because the same supernet training budget (48 GPU-hours for DAOS-A) yields both better accuracy and stabler selection.
  • The MMD regularization should generalize to any one-shot NAS pipeline with a labeled source and unlabeled target split, making the method modular as the paper states.
  • The searched architectures stay within a roughly 1700M FLOP budget, so edge deployment of pathology classifiers becomes plausible.
  • Higher validation–test correlation means the top-ranked architecture chosen during search is more likely to remain the best after retraining, reducing wasted retraining runs.

Reading between the lines

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

  • Beyond the paper's own evidence, the same validation split serving as MMD target and architecture-selection set means the 0.71 correlation likely overstates how well DAOS predicts performance on a truly unseen site; a multi-site or multi-scanner evaluation would be needed to confirm the mechanism.
  • NSDI's diversity measure is defined on operation choices alone and ignores weight-sharing interactions between candidate subnets, so its benefit may shrink when supernet ranking noise dominates.
  • A natural stress test the paper does not run is replacing MMD with a stronger alignment method such as adversarial domain discrimination to see if the validation-test correlation improves further.
  • The t-SNE and CAM evidence suggests feature alignment is the active ingredient, predicting the method should transfer to other stain-variant histology tasks using the same search space.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes DAOS, a one-shot neural architecture search framework for breast cancer histopathology image classification. DAOS combines two ideas: a Network Similarity Directed Initialization (NSDI) strategy that constrains the average pairwise similarity of the initial evolutionary-search population, and a Maximum Mean Discrepancy (MMD) domain-adaptation loss added to the supernet training objective to align training and validation feature distributions. The authors evaluate on the BRACS dataset with a ShuffleNetV2-based search space and report that DAOS improves top-1 F1 score over SPOS and random search, improves validation-to-test ranking correlation, and yields more clinically relevant class activation maps. The central claims are that NSDI stabilizes search and that domain adaptation improves supernet evaluation reliability.

Significance. If the empirical claims were fully controlled, the contribution would be modest but useful: the combination of MMD-based domain adaptation with one-shot supernet training is a sensible idea for pathology data with stain variation, and the NSDI heuristic is a practical way to diversify the initial population in evolutionary NAS. The paper's strengths include the explicit formulation of the MMD objective in Eq. (10), the 10-seed search experiments in Table 4, and the direct comparison of validation-test correlation across training schemes in Figure 6. However, the paper does not release code or trained models, and the central performance claims currently rest on uncontrolled comparisons and on a validation split that is used both as the adaptation target and as the ranking set. As a result, the demonstrated significance is substantially weaker than the abstract claims.

major comments (4)
  1. [§4.2.2 and Table 3] The training schedules are not matched across methods. The text states that the SPOS baseline is trained for 2000 epochs, DAOS-A adds 1000 epochs of MMD fine-tuning, and DAOS-B adds a further 1000 epochs of frozen-classifier fine-tuning, yet Table 3 reports identical 48 GPU-hours for SPOS and DAOS-A. This internal contradiction, together with the absence of an epoch-matched SPOS control (e.g., SPOS trained for 3000 epochs without MMD) and a DAOS-A variant without the MMD term, means that the F1 differences in Table 4 (61.41 vs 59.54 best; 59.28 vs 58.43 mean) and the correlation gains in §4.3.2 cannot be attributed to the proposed domain-adaptation loss.
  2. [§3.3, Eq. (10) and §4.3.2] The validation split is used as the unlabeled target distribution D_t during supernet training, and the same validation split is later used to rank architectures in the search (§4.2.3, Eq. (2)). The reported improvement in validation-test Pearson correlation from 0.1794 to 0.6985/0.7096 is therefore not an independent measure of ranking quality: the supernet has been explicitly trained to align its features to the validation split, making the validation metric a partially self-fulfilling ranking signal. The authors should either use a separate unlabeled set for the MMD target or report the correlation between validation-ranked and test-ranked architectures on an untouched holdout.
  3. [§4.3.3, Table 5] The statement that DAOS-A achieves the highest F1 score among all methods relies on Table 5, which reports no error bars or number of trials, while Table 4 reports 10-seed results whose mean and standard deviation overlap across methods (e.g., DAOS-A 59.28±0.97 vs SPOS 58.43±0.64; DAOS-B 59.35±0.73). Overlapping standard deviations do not establish a statistically significant improvement, and the 'Best' row is a best-of-seed statistic that inflates apparent gains. Please provide paired per-seed comparisons, confidence intervals, or significance tests, and report error bars in Table 5.
  4. [§4.3.1 and Table 2] The contribution of NSDI is not isolated from the domain-adaptation changes. Table 2 compares EA(NSDI) with EA(RI) only on validation F1 (mean 68.89 vs 68.37, std 0.69 vs 0.82) with no significance test and no test-set follow-up, while the final comparisons in Table 4 change both the initialization method (RI→NSDI) and the supernet objective (CL→CL+DAL) simultaneously. The paper needs a 2×2 ablation (initialization method × objective) on the final test metric to support the separate claims that NSDI stabilizes search and that domain adaptation improves evaluation reliability.
minor comments (5)
  1. [§3.3, Eq. (9)] Equation (9) uses the kernel function k(x, x') without specifying the kernel type or bandwidth; since this is a free parameter of the MMD estimator that directly enters Eq. (10), it should be reported.
  2. [§4.2.1] The text states that the search space contains '420 possible architectures'; this appears to be a typo for 4^20 (approximately 1.1×10^12), and the search-space size should be corrected.
  3. [Tables 4 and 5] Tables 4 and 5 use inconsistent reporting conventions: Table 4 gives mean±std over 10 seeds, while Table 5 gives only a single F1 value with no indication of which run or how many runs were used; this should be clarified.
  4. [General formatting] Repeated 'PRIME AI paper' headers appear between sections; if these are a typesetting artifact, they should be removed before publication.
  5. [Figure 2] The red-box annotation highlighting under-represented operators is difficult to discern in a black-and-white rendering; please use color or add explicit layer labels to make the point visible.

Circularity Check

1 steps flagged · score 2.0 of 10

One self-definitional NSDI diversity check; central performance claims are empirically benchmarked and not circular.

  1. self definitional [Section 3.1, Eq. (4), Algorithm 1; Section 4.3.1, Table 1]
    "We then define the Average Population Similarity (APS) as: APS(V P ) ≜ 1/P Σ_i max_{v_j ∈ V_P, i≠j} SS(v_i, v_j). ... To ensure diversity, we constrain APS under a user-defined threshold AP S_max ... By contrast, our NSDI strategy—with AP S_max = 6 and timeout T = 2 × 10^5—achieves lower APS scores, leading to better diversity."

    The APS metric used as evidence of NSDI's benefit is exactly the quantity Algorithm 1 is written to minimize: candidates are rejected whenever SS_max(v, V) > AP S_max (Algorithm 1, lines 7-8), with the timeout only ratcheting the threshold upward. Hence Table 1's lower APS for NSDI is logically entailed by the construction of NSDI, not an independently observed effect of it. This is a self-definitional confirmation. It does not, however, compromise the downstream claims: the search-performance gains in Table 2/Fig. 4 and the DAOS comparisons against SPOS are external empirical benchmarks, not derived from the APS definition.

full rationale

No load-bearing circularity found in the central derivation chain. The MMD objective (Eq. 10) is a standard two-sample discrepancy regularizer; the reported validation-test Pearson correlations (0.1794 vs 0.6985/0.7096) are not algebraically forced by Eq. (10) because the test set appears neither in the MMD term nor in the search objective. Baseline comparisons to SPOS, random search, and manual CNNs are external and epoch-matched claims, though Table 3's equal GPU-hour entry for SPOS and DAOS-A is internally inconsistent with the 3000-epoch DAOS-A schedule described in Section 4.2.2; that is a correctness/confounding concern, not a circularity. The only by-construction element is the NSDI/APS demonstration, where the diversity metric is the same quantity the initialization algorithm is designed to constrain. The validation set is used as both the MMD target and the architecture-ranking set, which is a methodological leakage concern, but it does not reduce any prediction to its input in the Eq-to-Eq sense. Score 2 reflects one minor self-definitional confirmation; the paper's main performance and correlation claims retain independent empirical content.

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

The paper's own performance gains depend on a chain of hand-set quantities: the MMD weight, the diversity threshold, the timeout, the training schedule, and the kernel bandwidth. None are externally benchmarked. The chain is evaluated on a single dataset.

free parameters (6)
  • MMD balancing coefficient lambda = 0.5
    Set in §3.3 Eq (10) without sensitivity analysis; it directly controls the strength of the domain-adaptation term that the paper credits for the improved validation-test correlation.
  • NSDI similarity threshold APS_max = 6
    Hand-set threshold in Algorithm 1 that defines how diverse the initial NAS population must be; no ablation is given, and Table 1 reports observed APS near 7 despite the threshold, indicating adaptive increments.
  • NSDI sampling timeout T = 200000
    Hand-set timeout in Algorithm 1 after which APS_max is raised; it interacts with latency constraints and affects the diversity of the resulting population.
  • MMD RBF kernel bandwidth = Not specified
    The kernel k(x,x') in Eq (8) requires a bandwidth; the paper never states it, so a key component of the domain-adaptation loss is underspecified.
  • Supernet training schedule (epochs) = 2000 baseline, 3000 DAOS-A, 4000 DAOS-B per §4.2.2
    These are chosen by hand and confound the central ablation: DAOS-A gets 1000 more epochs than the SPOS baseline, while Table 3 claims equal training time. The effect of MMD is therefore not isolated.
  • Latency/FLOPs bound Latmax = 1800M FLOPs
    Chosen in §4.2.3 because of the large image size; it defines the feasible search space and interacts with the NSDI acceptance criterion.
assumptions (6)
  • standard math The MMD estimator in Eq (9) is unbiased for the chosen kernel.
    Quoted from Gretton et al. [35] and accepted as background, but its usefulness depends on the unspecified kernel bandwidth.
  • domain assumption The BRACS training split and validation split form distinct source and target domains with a stain or scale shift that MMD can correct.
    Introduced in §3.3 and relied on by Eq (10); without a real domain shift, the MMD term is just an extra regularizer and the improved correlation is unexplained.
  • domain assumption Supernet weights trained by uniform random sampling under FLOPs constraints are good enough that inherited weights rank candidate architectures reliably.
    Standard one-shot assumption from [33]; used in §3.2 Eq (6) and §4.2.2. The paper's search ranking depends on it.
  • ad hoc to paper The XNOR-based similarity definition in Eq (3) accurately captures architectural diversity for the evolutionary search.
    The paper motivates it by analogy to force-directed placement but provides no validation that APS correlates with useful search progress; it is the basis of NSDI.
  • domain assumption BRACS is representative enough to support general conclusions about pathological image NAS.
    All experiments use this single dataset (§4.1); no cross-dataset test is performed.
  • domain assumption The FLOPs budget approximates clinical deployment constraints on edge devices.
    Used in §3.1 and §4.2.3; the paper does not verify on-device latency or clinical utility.

how reviews work

0 comments
Cite this review

Pith. "Pith review of One-Shot Neural Architecture Search with Network Similarity Directed Initialization for Pathological Image Classification." pith.science (2026). https://pith.science/paper/ZSOBCDUM

@misc{pith2026250614176,
  author       = {Pith},
  title        = {Pith review of: One-Shot Neural Architecture Search with Network Similarity Directed Initialization for Pathological Image Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZSOBCDUM}},
  note         = {Machine review of arXiv:2506.14176}
}
read the original abstract

Deep learning-based pathological image analysis presents unique challenges due to the practical constraints of network design. Most existing methods apply computer vision models directly to medical tasks, neglecting the distinct characteristics of pathological images. This mismatch often leads to computational inefficiencies, particularly in edge-computing scenarios. To address this, we propose a novel Network Similarity Directed Initialization (NSDI) strategy to improve the stability of neural architecture search (NAS). Furthermore, we introduce domain adaptation into one-shot NAS to better handle variations in staining and semantic scale across pathology datasets. Experiments on the BRACS dataset demonstrate that our method outperforms existing approaches, delivering both superior classification performance and clinically relevant feature localization.

Figures

Figures reproduced from arXiv: 2506.14176 by the authors.

Figure 1
Figure 1. The overview of the proposed domain adaptation one-shot neural architecture search. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. One sample of population initialization under a specific random seed, where the population number is 50, and [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. The impact of population initialization on the evolutionary algorithm. (a)Random sampling, (b)mutation and [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Comparison of random search and evolutionary algorithm with random initialization or network similarity [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Best architecture searched by DAOS. 4.3.2 Domain Adaptation Improves Supernet Training We compare three supernet training schemes: • Baseline: 2000 epochs of training using classification loss only. • DAOS-A: Initial 2000 epochs with classification loss, followed by 10…
Figure 6
Figure 6. Figure 6: Results of 1000 network evaluation metrics on the validation set and test set. All networks were randomly [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: t-SNE of the average supernet features on the training set by different methods. (a) Baseline, (b) DAOS-A, [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 8
Figure 8. Figure 8: CAMs generated from different methods. The upper left corner of the images show the real labels or predicted [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

46 extracted references · 39 canonical work pages

  1. [1]

    Global cancer statistics 2020: Globocan estimates of incidence and mortality worldwide for 36 cancers in 185 countries

    Hyuna Sung, Jacques Ferlay, Rebecca L Siegel, Mathieu Laversanne, Isabelle Soerjomataram, Ahmedin Jemal, and Freddie Bray. Global cancer statistics 2020: Globocan estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA: a cancer journal for clinicians, 71(3):209–249, 2021

    work page 2020

  2. [2]

    Risk factors and preventions of breast cancer

    Yi-Sheng Sun, Zhao Zhao, Zhang-Nv Yang, Fang Xu, Hang-Jing Lu, Zhi-Yong Zhu, Wen Shi, Jianmin Jiang, Ping-Ping Yao, and Han-Ping Zhu. Risk factors and preventions of breast cancer. International journal of biological sciences, 13(11):1387, 2017

    work page 2017

  3. [3]

    Global surveillance of cancer survival 1995–2009: analysis of individual data for 25 676 887 patients from 279 population-based registries in 67 countries (concord-2)

    Claudia Allemani, Hannah K Weir, Helena Carreira, et al. Global surveillance of cancer survival 1995–2009: analysis of individual data for 25 676 887 patients from 279 population-based registries in 67 countries (concord-2). The Lancet, 385(9972):977–1010, 2015. 11 PRIME AI paper

    work page 1995

  4. [4]

    Douglas S Gomes, Simone S Porto, Débora Balabram, and Helenice Gobbi. Inter-observer variability between general pathologists and a specialist in breast pathology in the diagnosis of lobular neoplasia, columnar cell lesions, atypical ductal hyperplasia and ductal carcinoma in situ of the breast. Diagnostic pathology, 9:1–9, 2014

    work page 2014

  5. [5]

    Elmore, Gary M

    Joann G. Elmore, Gary M. Longton, Patricia A. Carney, Berta M. Geller, Tracy Onega, Anna N. A. Tosteson, Heidi D. Nelson, Margaret S. Pepe, Kimberly H. Allison, Stuart J. Schnitt, Frances P. O’Malley, and Donald L. Weaver. Diagnostic Concordance Among Pathologists Interpreting Breast Biopsy Specimens.JAMA, 313(11):1122– 1132, 03 2015

    work page 2015

  6. [6]

    Neural architecture search with reinforcement learning

    Barret Zoph and Quoc V Le. Neural architecture search with reinforcement learning. arXiv preprint arXiv:1611.01578, 2016

    arXiv 2016

  7. [7]

    Learning transferable architectures for scalable image recognition

    Barret Zoph, Vijay Vasudevan, Jonathon Shlens, and Quoc V Le. Learning transferable architectures for scalable image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 8697–8710, 2018

    work page 2018

  8. [8]

    Darts: Differentiable architecture search

    Hanxiao Liu, Karen Simonyan, and Yiming Yang. Darts: Differentiable architecture search. arXiv preprint arXiv:1806.09055, 2018

    arXiv 2018

Show all 46 references
  1. [9]

    Proxylessnas: Direct neural architecture search on target task and hardware

    Han Cai, Ligeng Zhu, and Song Han. Proxylessnas: Direct neural architecture search on target task and hardware. arXiv preprint arXiv:1812.00332, 2018

    2018 arXiv

  2. [10]

    Multinomial distribution learning for effective neural architecture search

    Xiawu Zheng, Rongrong Ji, Lang Tang, Baochang Zhang, Jianzhuang Liu, and Qi Tian. Multinomial distribution learning for effective neural architecture search. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 1304–1313, 2019

    2019

  3. [11]

    Mnasnet: Platform-aware neural architecture search for mobile

    Mingxing Tan, Bo Chen, Ruoming Pang, Vijay Vasudevan, Mark Sandler, Andrew Howard, and Quoc V Le. Mnasnet: Platform-aware neural architecture search for mobile. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 2820–2828, 2019

    2019

  4. [12]

    Fbnet: Hardware-aware efficient convnet design via differentiable neural architecture search

    Bichen Wu, Xiaoliang Dai, Peizhao Zhang, Yanghan Wang, Fei Sun, Yiming Wu, Yuandong Tian, Peter Vajda, Yangqing Jia, and Kurt Keutzer. Fbnet: Hardware-aware efficient convnet design via differentiable neural architecture search. In Proceedings of the IEEE/CVF Conference on Com...

    2019

  5. [13]

    Fbnetv2: Differentiable neural architecture search for spatial and channel dimensions

    Alvin Wan, Xiaoliang Dai, Peizhao Zhang, Zijian He, Yuandong Tian, Saining Xie, Bichen Wu, Matthew Yu, Tao Xu, Kan Chen, et al. Fbnetv2: Differentiable neural architecture search for spatial and channel dimensions. In Proceedings of the IEEE/CVF Conference on Computer Vision a...

    2020

  6. [14]

    Regularized evolution for image classifier architecture search

    Esteban Real, Alok Aggarwal, Yanping Huang, and Quoc V Le. Regularized evolution for image classifier architecture search. In Proceedings of the aaai conference on artificial intelligence, volume 33, pages 4780–4789, 2019

    2019

  7. [15]

    Learning curve prediction with bayesian neural networks

    Aaron Klein, Stefan Falkner, Jost Tobias Springenberg, and Frank Hutter. Learning curve prediction with bayesian neural networks. In International Conference on Learning Representations, 2017

    2017

  8. [16]

    A downsampled variant of imagenet as an alternative to the cifar datasets

    Patryk Chrabaszcz, Ilya Loshchilov, and Frank Hutter. A downsampled variant of imagenet as an alternative to the cifar datasets. arXiv preprint arXiv:1707.08819, 2017

    2017 arXiv

  9. [17]

    Speeding up automatic hyperparameter optimization of deep neural networks by extrapolation of learning curves

    Tobias Domhan, Jost Tobias Springenberg, and Frank Hutter. Speeding up automatic hyperparameter optimization of deep neural networks by extrapolation of learning curves. In Twenty-fourth international joint conference on artificial intelligence, 2015

    2015

  10. [18]

    Efficient architecture search by network transformation

    Han Cai, Tianyao Chen, Weinan Zhang, Yong Yu, and Jun Wang. Efficient architecture search by network transformation. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 32, 2018

    2018

  11. [19]

    Nas-unet: Neural architecture search for medical image segmentation

    Yu Weng, Tianbao Zhou, Yujie Li, and Xiaoyu Qiu. Nas-unet: Neural architecture search for medical image segmentation. IEEE access, 7:44247–44257, 2019

    2019

  12. [20]

    Probeable darts with application to computational pathology

    Sheyang Tang, Mahdi S Hosseini, Lina Chen, Sonal Varma, Corwyn Rowsell, Savvas Damaskinos, Konstantinos N Plataniotis, and Zhou Wang. Probeable darts with application to computational pathology. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages ...

    2021

  13. [21]

    Plexusnet: A neural network architectural concept for medical image classification

    Okyaz Eminaga, Mahmoud Abbas, Jeanne Shen, Mark Laurie, James D Brooks, Joseph C Liao, and Daniel L Rubin. Plexusnet: A neural network architectural concept for medical image classification. Computers in Biology and Medicine, page 106594, 2023

    2023

  14. [22]

    Ms-nas: Multi-scale neural architecture search for medical image segmentation

    Xingang Yan, Weiwen Jiang, Yiyu Shi, and Cheng Zhuo. Ms-nas: Multi-scale neural architecture search for medical image segmentation. In Medical Image Computing and Computer Assisted Intervention–MICCAI 2020: 23rd International Conference, Lima, Peru, October 4–8, 2020, Proceedi...

    2020

  15. [23]

    Neural architecture search for adversarial medical image segmentation

    Nanqing Dong, Min Xu, Xiaodan Liang, Yiliang Jiang, Wei Dai, and Eric Xing. Neural architecture search for adversarial medical image segmentation. In Medical Image Computing and Computer Assisted Intervention– MICCAI 2019: 22nd International Conference, Shenzhen, China, Octobe...

    2019

  16. [24]

    Adwu-net: adaptive depth and width u-net for medical image segmentation by differentiable neural architecture search

    Ziyan Huang, Zehua Wang, Zhikai Yang, and Lixu Gu. Adwu-net: adaptive depth and width u-net for medical image segmentation by differentiable neural architecture search. In International Conference on Medical Imaging with Deep Learning, pages 576–589. PMLR, 2022

    2022

  17. [25]

    Differentiable zooming for multiple instance learning on whole-slide images

    Kevin Thandiackal, Boqi Chen, Pushpak Pati, Guillaume Jaume, Drew FK Williamson, Maria Gabrani, and Orcun Goksel. Differentiable zooming for multiple instance learning on whole-slide images. In Computer Vision–ECCV 2022: 17th European Conference, Tel Aviv, Israel, October 23–2...

    2022

  18. [26]

    Stain-invariant self supervised learning for histopathology image analysis

    Alexandre Tiard, Alex Wong, David Joon Ho, Yangchao Wu, Eliram Nof, Stefano Soatto, and Saad Nadeem. Stain-invariant self supervised learning for histopathology image analysis. arXiv preprint arXiv:2211.07590, 2022

    2022 arXiv

  19. [27]

    Spatial-hierarchical graph neural network with dynamic structure learning for histological image classification

    Wentai Hou, Helong Huang, Qiong Peng, Rongshan Yu, Lequan Yu, and Liansheng Wang. Spatial-hierarchical graph neural network with dynamic structure learning for histological image classification. In Medical Image Computing and Computer Assisted Intervention–MICCAI 2022: 25th In...

    2022

  20. [28]

    Breast tumor image classification in bright challenge via multiple instance learning and deep transformers

    Yangen Zhan, Hao Bian, Yang Chen, Xiu Li, and Yongbing Zhang. Breast tumor image classification in bright challenge via multiple instance learning and deep transformers. In 2022 IEEE International Symposium on Biomedical Imaging Challenges (ISBIC), pages 1–5. IEEE, 2022

    2022

  21. [29]

    Scl-wc: Cross-slide contrastive learning for weakly-supervised whole-slide image classification

    Xiyue Wang, Jinxi Xiang, Jun Zhang, Sen Yang, Zhongyi Yang, Ming-Hui Wang, Jing Zhang, Yang Wei, Junzhou Huang, and Xiao Han. Scl-wc: Cross-slide contrastive learning for weakly-supervised whole-slide image classification. In Advances in Neural Information Processing Systems

  22. [30]

    A multi-task multiple instance learning algorithm to analyze large whole slide images from bright challenge 2022

    Niccolò Marini, Marek Wodzinski, Manfredo Atzori, and Henning Müller. A multi-task multiple instance learning algorithm to analyze large whole slide images from bright challenge 2022. In 2022 IEEE International Symposium on Biomedical Imaging Challenges (ISBIC), pages 1–4. IEEE, 2022

    2022

  23. [31]

    Multiple-instance learning with efficient transformer for breast tumor image classification in bright challenge

    Feng Wentai, Kuang Jinbo, Ji Zheng, and Xu Shuoyu. Multiple-instance learning with efficient transformer for breast tumor image classification in bright challenge. In 2022 IEEE International Symposium on Biomedical Imaging Challenges (ISBIC), pages 1–5. IEEE, 2022

    2022

  24. [32]

    Hierarchical graph representations in digital pathology

    Pushpak Pati, Guillaume Jaume, Antonio Foncubierta-Rodriguez, Florinda Feroce, Anna Maria Anniciello, Giosue Scognamiglio, Nadia Brancati, Maryse Fiche, Estelle Dubruc, Daniel Riccio, et al. Hierarchical graph representations in digital pathology. Medical image analysis, 75:10...

    2022

  25. [33]

    Single path one-shot neural architecture search with uniform sampling

    Zichao Guo, Xiangyu Zhang, Haoyuan Mu, Wen Heng, Zechun Liu, Yichen Wei, and Jian Sun. Single path one-shot neural architecture search with uniform sampling. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XVI 16, page...

    2020

  26. [34]

    Graph drawing by force-directed placement

    Thomas MJ Fruchterman and Edward M Reingold. Graph drawing by force-directed placement. Software: Practice and experience, 21(11):1129–1164, 1991

    1991

  27. [35]

    A kernel two-sample test

    Arthur Gretton, Karsten M Borgwardt, Malte J Rasch, Bernhard Schölkopf, and Alexander Smola. A kernel two-sample test. The Journal of Machine Learning Research, 13(1):723–773, 2012

    2012

  28. [36]

    Bracs: A dataset for breast carcinoma subtyping in h&e histology images

    Nadia Brancati, Anna Maria Anniciello, Pushpak Pati, Daniel Riccio, Giosuè Scognamiglio, Guillaume Jaume, Giuseppe De Pietro, Maurizio Di Bonito, Antonio Foncubierta, Gerardo Botti, et al. Bracs: A dataset for breast carcinoma subtyping in h&e histology images. Database, 2022, 2022

    2022

  29. [37]

    Training data-efficient image transformers & distillation through attention

    Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, and Herve Jegou. Training data-efficient image transformers & distillation through attention. In International Conference on Machine Learning, volume 139, pages 10347–10357, July 2021

    2021

  30. [38]

    Autobss: An efficient algorithm for block stacking style search

    Yikang Zhang, Jian Zhang, and Zhao Zhong. Autobss: An efficient algorithm for block stacking style search. Advances in Neural Information Processing Systems, 33:10259–10270, 2020

    2020

  31. [39]

    Learning deep features for discriminative localization

    Bolei Zhou, Aditya Khosla, Agata Lapedriza, Aude Oliva, and Antonio Torralba. Learning deep features for discriminative localization. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 2921–2929, 2016

    2016

  32. [40]

    An accurate prediction of the origin for bone metastatic cancer using deep learning on digital pathological images

    Lianghui Zhu, Huijuan Shi, Huiting Wei, Chengjiang Wang, Shanshan Shi, Fenfen Zhang, Renao Yan, Yiqing Liu, Tingting He, Liyuan Wang, et al. An accurate prediction of the origin for bone metastatic cancer using deep learning on digital pathological images. EBioMedicine, 87, 20...

    2023

  33. [41]

    Unpaired virtual histological staining using prior-guided generative adversarial networks

    Renao Yan, Qiming He, Yiqing Liu, Peng Ye, Lianghui Zhu, Shanshan Shi, Jizhou Gou, Yonghong He, Tian Guan, and Guangde Zhou. Unpaired virtual histological staining using prior-guided generative adversarial networks. Computerized Medical Imaging and Graphics, 105:102185, 2023

    2023

  34. [42]

    Shapley values-enabled progressive pseudo bag augmentation for whole-slide image classification.IEEE Transactions on Medical Imaging, 2024

    Renao Yan, Qiehe Sun, Cheng Jin, Yiqing Liu, Yonghong He, Tian Guan, and Hao Chen. Shapley values-enabled progressive pseudo bag augmentation for whole-slide image classification.IEEE Transactions on Medical Imaging, 2024

    2024

  35. [43]

    Texq: zero-shot network quantization with texture feature distribution calibration

    Xinrui Chen, Yizhi Wang, Renao Yan, Yiqing Liu, Tian Guan, and Yonghong He. Texq: zero-shot network quantization with texture feature distribution calibration. Advances in Neural Information Processing Systems, 36:274–287, 2023

    2023

  36. [44]

    Mergeup-augmented semi-weakly supervised learning for wsi classification

    Mingxi Ouyang, Yuqiu Fu, Renao Yan, ShanShan Shi, Xitong Ling, Lianghui Zhu, Yonghong He, and Tian Guan. Mergeup-augmented semi-weakly supervised learning for wsi classification. arXiv preprint arXiv:2408.12825, 2024

    2024 arXiv

  37. [45]

    Task-oriented embedding counts: Heuristic clustering-driven feature fine-tuning for whole slide image classification

    Xuenian Wang, Shanshan Shi, Renao Yan, Qiehe Sun, Lianghui Zhu, Tian Guan, and Yonghong He. Task-oriented embedding counts: Heuristic clustering-driven feature fine-tuning for whole slide image classification. arXiv preprint arXiv:2406.00672, 2024

    2024 arXiv

  38. [46]

    Hierarchically optimized multiple instance learning with multi-magnification pathological images for cerebral tumor diagnosis

    Lianghui Zhu, Renao Yan, Tian Guan, Fenfen Zhang, Linlang Guo, Qiming He, Shanshan Shi, Huijuan Shi, Yonghong He, and Anjia Han. Hierarchically optimized multiple instance learning with multi-magnification pathological images for cerebral tumor diagnosis. IEEE Journal of Biome...

    2025

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.