pith. sign in

arxiv: 2604.17254 · v1 · submitted 2026-04-19 · 📊 stat.ME · stat.AP

Detecting Breast Carcinoma Metastasis on Whole-Slide Images by Partially Subsampled Multiple Instance Learning

Baichen Yu , Xuetong Li , Jing Zhou , Hansheng Wang This is my paper

Pith reviewed 2026-05-10 06:21 UTC · model grok-4.3

classification 📊 stat.ME stat.AP
keywords breast cancer metastasiswhole-slide imagingmultiple instance learningGaussian mixture modelmaximum likelihood estimationhistopathology analysis
0
0 comments X

The pith

A Gaussian mixture multiple instance learning framework with partial subsampling detects breast cancer metastases in whole-slide images more accurately than prior methods.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes a Gaussian mixture model for multiple instance learning on whole-slide histopathology images to identify breast carcinoma metastases. It proposes a bag-based maximum likelihood estimator that predicts at the slide level from randomly cropped sub-images treated as instances, along with a subsampling-based estimator that refines accuracy by labeling only selected instances. Evaluations on breast cancer metastasis prediction show the bag-based estimator exceeds existing approaches while the subsampling version improves performance at both slide and sub-image levels. The method is presented as robust to plausible model errors, with supporting theory and simulations.

Core claim

Each whole-slide image is modeled as a bag of instances consisting of randomly cropped sub-images assumed to follow a Gaussian mixture distribution. A bag-based maximum likelihood estimator predicts metastasis presence from the bag, and a subsampling-based maximum likelihood estimator improves predictions by selectively labeling a subset of instances. On breast carcinoma metastasis tasks, the bag-based estimator surpasses state-of-the-art methods and the subsampling estimator further raises accuracy at bag and instance levels, with robustness to model mis-specifications demonstrated through theory and simulations.

What carries the argument

The Gaussian mixture multiple instance learning setup, where the bag-based maximum likelihood estimator aggregates instance-level probabilities across randomly cropped sub-images and the subsampling-based maximum likelihood estimator refines them through selective labeling to handle large image sizes and tissue heterogeneity.

If this is right

  • Prediction accuracy rises at the whole-slide level and at the individual sub-image level compared with prior multiple instance learning techniques.
  • The approach remains effective even when the Gaussian mixture assumption is imperfectly met due to tissue heterogeneity.
  • Computational demands decrease because only a subset of sub-images needs labeling while still using the full bag structure for initial estimation.
  • Theoretical consistency of the estimators supports reliable deployment in clinical pathology workflows.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The framework could apply to metastasis detection in other cancer types by retraining the mixture components on domain-specific image crops.
  • Replacing raw pixel instances with deep-learned features inside the same maximum likelihood structure might yield further gains without changing the overall estimators.
  • Robustness to mis-specification suggests the method could tolerate the label noise typical in large medical image collections.

Load-bearing premise

Randomly cropped sub-images from whole-slide images follow a Gaussian mixture distribution whose parameters can be estimated reliably by maximum likelihood without major bias introduced by partial subsampling or tissue variations.

What would settle it

An independent test set of whole-slide images where the subsampling-based estimator fails to improve both bag-level and instance-level accuracy over the bag-based estimator or standard multiple instance learning baselines would show the claimed gains do not hold.

read the original abstract

Breast cancer is the most prevalent cancer in women worldwide. Histopathology image analysis serves as the gold standard for cancer diagnosis. In this regard, whole-slide imaging (WSI), a revolutionary technology in digital pathology, allows for ultrahigh-resolution tissue analysis. Despite its promise, WSI analysis faces significant computational challenges due to its massive data size and tissue heterogeneity. To address this issue, we present a Gaussian mixture based multiple instance learning (MIL) framework for WSI analysis with partially subsampled instances. Our approach models a WSI as a bag of instances (i.e., randomly cropped sub-images), leveraging a bag-based maximum likelihood estimator (BMLE) to predict metastases. Furthermore, we introduce a subsampling-based maximum likelihood estimator (SMLE) to refine predictions by selectively labeling a subset of instances. Extensive evaluations of the breast carcinoma metastasis prediction demonstrate that BMLE surpasses state-of-the-art methods, while the SMLE further improves the prediction accuracy at both bag and instance levels. We find that our method is fairly robust against various plausible model mis-specifications. Theoretical analyses and simulation studies validate the performance and robustness of our methods.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper proposes a Gaussian mixture model (GMM)-based multiple instance learning (MIL) framework for detecting breast carcinoma metastasis on whole-slide images (WSIs). WSIs are modeled as bags of randomly cropped sub-image instances assumed to be i.i.d. draws from a GMM; a bag-based maximum likelihood estimator (BMLE) is used for metastasis prediction, and a subsampling-based MLE (SMLE) is introduced to refine predictions by selectively labeling instances. The central claims are that BMLE surpasses state-of-the-art methods, SMLE further improves accuracy at both bag and instance levels, and the approach is robust to plausible model mis-specifications, with support from theoretical analyses and simulation studies.

Significance. If the empirical claims and robustness hold under realistic conditions, the work offers a statistically principled, likelihood-based alternative to deep-learning MIL methods for large-scale WSI analysis. The use of partial subsampling to manage computational burden and the focus on deriving estimators from maximum likelihood principles are strengths that could improve interpretability and reliability in digital pathology applications.

major comments (2)
  1. [Abstract] Abstract: the claims that 'BMLE surpasses state-of-the-art methods' and 'SMLE further improves the prediction accuracy' are load-bearing for the paper's contribution yet are stated without any quantitative metrics, error bars, dataset sizes, or baseline comparisons, preventing assessment of effect sizes or statistical significance.
  2. [Method and robustness analysis] Method (GMM for instances) and robustness section: the framework treats randomly cropped instances as i.i.d. from a GMM whose parameters are estimated by BMLE or SMLE; however, WSIs exhibit spatial correlations, varying cellular densities, and staining artifacts that violate both the Gaussian component shape and the i.i.d. assumption. The reported robustness is validated only via simulations that presumably generate data from the same GMM family; no real-data ablation under controlled departures from the assumed distribution is described, which directly undermines the consistency and superiority claims on actual slides.
minor comments (2)
  1. [Abstract] Abstract: the phrase 'extensive evaluations of the breast carcinoma metastasis prediction' is vague; it should specify the number of WSIs, patients, cross-validation folds, and exact performance measures used.
  2. [Method] Notation: the distinction between bag-level and instance-level predictions under partial subsampling should be clarified with explicit equations for the likelihood contributions of labeled versus unlabeled instances.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments, which have helped clarify the presentation of our work. We respond point by point below and have revised the manuscript to address the concerns.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the claims that 'BMLE surpasses state-of-the-art methods' and 'SMLE further improves the prediction accuracy' are load-bearing for the paper's contribution yet are stated without any quantitative metrics, error bars, dataset sizes, or baseline comparisons, preventing assessment of effect sizes or statistical significance.

    Authors: We agree that the abstract would be strengthened by including quantitative details. In the revised manuscript we have added specific performance metrics (including means and standard deviations across cross-validation folds), the number of WSIs and instances used, and explicit numerical comparisons against the baselines reported in the experiments section. revision: yes

  2. Referee: [Method and robustness analysis] Method (GMM for instances) and robustness section: the framework treats randomly cropped instances as i.i.d. from a GMM whose parameters are estimated by BMLE or SMLE; however, WSIs exhibit spatial correlations, varying cellular densities, and staining artifacts that violate both the Gaussian component shape and the i.i.d. assumption. The reported robustness is validated only via simulations that presumably generate data from the same GMM family; no real-data ablation under controlled departures from the assumed distribution is described, which directly undermines the consistency and superiority claims on actual slides.

    Authors: The referee correctly notes that the i.i.d. Gaussian-mixture assumption is an approximation; real WSIs contain spatial structure and staining variation. Our theoretical results establish consistency when the model is correctly specified, while the simulation studies deliberately introduce controlled departures (non-Gaussian components, varying mixture weights, and weak dependence) to probe robustness. The real-data experiments show that BMLE and SMLE still outperform competing methods, which we interpret as evidence of practical utility under model mismatch. We have revised the robustness section to state these limitations more explicitly, to describe the simulation designs in greater detail, and to qualify the scope of the robustness claims. We have not added new controlled real-data ablation experiments. revision: partial

Circularity Check

0 steps flagged

No circularity: standard MLE derivation on explicit GMM-MIL model

full rationale

The paper defines a Gaussian mixture model for randomly cropped WSI instances and applies standard maximum likelihood estimation to obtain BMLE and SMLE. These estimators follow directly from the likelihood function under the stated model assumptions without any reduction to fitted parameters by construction, self-citation load-bearing premises, or renaming of known results. Theoretical analyses and simulations are presented as validation steps separate from the core derivation. No load-bearing step equates a claimed prediction or uniqueness result to its own inputs; the framework remains self-contained against external statistical benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

Based on abstract only; the framework rests on standard statistical assumptions for MIL and MLE plus the domain-specific choice of Gaussian mixtures for instance modeling. No free parameters or invented entities explicitly named.

axioms (2)
  • domain assumption WSI instances can be modeled as draws from a Gaussian mixture distribution
    Core modeling choice stated in the abstract for the bag of instances.
  • domain assumption Maximum likelihood estimation yields reliable bag-level predictions under partial subsampling
    Underpins both BMLE and SMLE performance claims.

pith-pipeline@v0.9.0 · 5507 in / 1261 out tokens · 38276 ms · 2026-05-10T06:21:04.588628+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

171 extracted references · 171 canonical work pages · 2 internal anchors

  1. [1]

    IEEE Transactions on Cybernetics , volume=

    Weakly supervised deep learning for whole slide lung cancer image analysis , author=. IEEE Transactions on Cybernetics , volume=. 2019 , publisher=

    work page 2019

  2. [2]

    2003 , publisher=

    An Introduction to Multivariate Statistical Analysis , author=. 2003 , publisher=

    work page 2003

  3. [3]

    Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages=

    Patch-based convolutional neural network for whole slide tissue image classification , author=. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages=

    work page

  4. [4]

    Big Data II: Learning, Analytics, and Applications , volume=

    Large-scale storage of whole slide images and fast retrieval of tiles using DRAM , author=. Big Data II: Learning, Analytics, and Applications , volume=. 2020 , organization=

    work page 2020

  5. [5]

    Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages=

    Thoracic disease identification and localization with limited supervision , author=. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages=

    work page

  6. [6]

    Annual Review of Pathology: Mechanisms of Disease , volume=

    Digital imaging in pathology: whole-slide imaging and beyond , author=. Annual Review of Pathology: Mechanisms of Disease , volume=. 2013 , publisher=

    work page 2013

  7. [7]

    Proceedings of the AMIA Annual Fall Symposium , pages=

    The virtual microscope , author=. Proceedings of the AMIA Annual Fall Symposium , pages=. 1997 , organization=

    work page 1997

  8. [8]

    2022 IEEE International Symposium on Biomedical Imaging Challenges (ISBIC) , pages=

    A multi-task multiple instance learning algorithm to analyze large whole slide images from bright challenge 2022 , author=. 2022 IEEE International Symposium on Biomedical Imaging Challenges (ISBIC) , pages=. 2022 , organization=

    work page 2022

  9. [9]

    2020 , publisher=

    Sampling Theory and Practice , author=. 2020 , publisher=

    work page 2020

  10. [10]

    2003 , publisher=

    Mathematical Statistics , author=. 2003 , publisher=

    work page 2003

  11. [11]

    2019 , publisher=

    Matrix Differential Calculus with Applications in Statistics and Econometrics , author=. 2019 , publisher=

    work page 2019

  12. [12]

    Journal of the American Statistical Association , volume=

    Maximum likelihood estimation of the multivariate normal mixture model , author=. Journal of the American Statistical Association , volume=. 2009 , publisher=

    work page 2009

  13. [13]

    2000 , publisher=

    Asymptotic Statistics , author=. 2000 , publisher=

    work page 2000

  14. [14]

    2019 , publisher=

    Probability: Theory and Examples , author=. 2019 , publisher=

    work page 2019

  15. [15]

    2002 , publisher=

    Applied Multivariate Statistical Analysis , author=. 2002 , publisher=

    work page 2002

  16. [16]

    Journal of the American Statistical Association , volume=

    Variable selection via nonconcave penalized likelihood and its oracle properties , author=. Journal of the American Statistical Association , volume=. 2001 , publisher=

    work page 2001

  17. [17]

    Econometrica , volume=

    Linear regression limit theory for nonstationary panel data , author=. Econometrica , volume=. 1999 , publisher=

    work page 1999

  18. [18]

    1968 , publisher=

    Real Analysis , author=. 1968 , publisher=

    work page 1968

  19. [19]

    NPJ Digital Medicine , volume=

    An ensemble deep learning model for risk stratification of invasive lung adenocarcinoma using thin-slice CT , author=. NPJ Digital Medicine , volume=. 2023 , publisher=

    work page 2023

  20. [20]

    Nature Communications , volume=

    Privacy risks of whole-slide image sharing in digital pathology , author=. Nature Communications , volume=. 2023 , publisher=

    work page 2023

  21. [21]

    Journal of Pathology Informatics , volume=

    Implementation of whole slide imaging in surgical pathology: A value added approach , author=. Journal of Pathology Informatics , volume=. 2011 , publisher=

    work page 2011

  22. [22]

    Pathology and Laboratory Medicine International , pages=

    Whole slide imaging in pathology: advantages, limitations, and emerging perspectives , author=. Pathology and Laboratory Medicine International , pages=. 2015 , publisher=

    work page 2015

  23. [23]

    Journal of Digital Imaging , volume=

    Whole slide imaging (WSI) in pathology: current perspectives and future directions , author=. Journal of Digital Imaging , volume=. 2020 , publisher=

    work page 2020

  24. [24]

    IEEE Access , volume=

    The devil is in the details: Whole slide image acquisition and processing for artifacts detection, color variation, and data augmentation: A review , author=. IEEE Access , volume=. 2022 , publisher=

    work page 2022

  25. [25]

    arXiv preprint arXiv:1606.05718 , year=

    Deep learning for identifying metastatic breast cancer , author=. arXiv preprint arXiv:1606.05718 , year=

    work page arXiv

  26. [26]

    IEEE Transactions on Systems, Man, and Cybernetics , volume=

    A threshold selection method from gray-level histograms , author=. IEEE Transactions on Systems, Man, and Cybernetics , volume=. 1979 , publisher=

    work page 1979

  27. [27]

    Drug Safety Evaluation: Methods and Protocols , pages=

    Histopathology procedures: from tissue sampling to histopathological evaluation , author=. Drug Safety Evaluation: Methods and Protocols , pages=. 2011 , publisher=

    work page 2011

  28. [28]

    JAMA , volume=

    Diagnostic assessment of deep learning algorithms for detection of lymph node metastases in women with breast cancer , author=. JAMA , volume=. 2017 , publisher=

    work page 2017

  29. [29]

    Very Deep Convolutional Networks for Large-Scale Image Recognition

    Very deep convolutional networks for large-scale image recognition , author=. arXiv preprint arXiv:1409.1556 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  30. [30]

    Proceedings of the European Conference on Computer Vision (ECCV) Workshops , year=

    A fast and scalable pipeline for stain normalization of whole-slide images in histopathology , author=. Proceedings of the European Conference on Computer Vision (ECCV) Workshops , year=

    work page

  31. [31]

    Expert Systems with Applications , volume=

    Multiple instance learning for histopathological breast cancer image classification , author=. Expert Systems with Applications , volume=. 2019 , publisher=

    work page 2019

  32. [32]

    IEEE Transactions on Medical Imaging , volume=

    Multi-instance multi-label learning for multi-class classification of whole slide breast histopathology images , author=. IEEE Transactions on Medical Imaging , volume=. 2017 , publisher=

    work page 2017

  33. [33]

    Nature Medicine , volume=

    Clinical-grade computational pathology using weakly supervised deep learning on whole slide images , author=. Nature Medicine , volume=. 2019 , publisher=

    work page 2019

  34. [34]

    National Science Review , volume=

    A brief introduction to weakly supervised learning , author=. National Science Review , volume=. 2018 , publisher=

    work page 2018

  35. [35]

    arXiv preprint arXiv:2304.06292 , year=

    Improved naive Bayes with mislabeled data , author=. arXiv preprint arXiv:2304.06292 , year=

    work page arXiv

  36. [36]

    Journal of the American Statistical Association , volume=

    Semi-supervised linear regression , author=. Journal of the American Statistical Association , volume=. 2022 , publisher=

    work page 2022

  37. [37]

    Big Data and Information Analytics , volume=

    Multiple-instance learning for text categorization based on semantic representation , author=. Big Data and Information Analytics , volume=

    work page

  38. [38]

    Using customer service dialogues for satisfaction analysis with context-assisted multiple instance learning , author=. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP) , pages=

    work page 2019

  39. [39]

    IEEE Transactions on Image Processing , volume=

    A multiple-instance densely-connected ConvNet for aerial scene classification , author=. IEEE Transactions on Image Processing , volume=. 2020 , publisher=

    work page 2020

  40. [40]

    IEEE Transactions on Geoscience and Remote Sensing , volume=

    Deep multiple instance convolutional neural networks for learning robust scene representations , author=. IEEE Transactions on Geoscience and Remote Sensing , volume=. 2020 , publisher=

    work page 2020

  41. [41]

    Journal of the Royal Statistical Society Series A: Statistics in Society , volume=

    Generalized linear models , author=. Journal of the Royal Statistical Society Series A: Statistics in Society , volume=. 1972 , publisher=

    work page 1972

  42. [42]

    Machine Learning , volume=

    Random forests , author=. Machine Learning , volume=. 2001 , publisher=

    work page 2001

  43. [43]

    Nature , volume=

    Deep learning , author=. Nature , volume=. 2015 , publisher=

    work page 2015

  44. [44]

    IEEE Transactions on Medical Imaging , volume=

    Fast scannet: Fast and dense analysis of multi-gigapixel whole-slide images for cancer metastasis detection , author=. IEEE Transactions on Medical Imaging , volume=. 2019 , publisher=

    work page 2019

  45. [45]

    Nature Communications , volume=

    An annotation-free whole-slide training approach to pathological classification of lung cancer types using deep learning , author=. Nature Communications , volume=. 2021 , publisher=

    work page 2021

  46. [46]

    Gastric Cancer , volume=

    Deep learning system for lymph node quantification and metastatic cancer identification from whole-slide pathology images , author=. Gastric Cancer , volume=. 2021 , publisher=

    work page 2021

  47. [47]

    Advances in Neural Information Processing Systems , volume=

    A framework for multiple-instance learning , author=. Advances in Neural Information Processing Systems , volume=

    work page

  48. [48]

    Advances in Neural Information Processing Systems , volume=

    Support vector machines for multiple-instance learning , author=. Advances in Neural Information Processing Systems , volume=

    work page

  49. [49]

    Pacific-Asia Conference on Knowledge Discovery and Data Mining , pages=

    Logistic regression and boosting for labeled bags of instances , author=. Pacific-Asia Conference on Knowledge Discovery and Data Mining , pages=. 2004 , organization=

    work page 2004

  50. [50]

    AAAI , pages=

    Instance-level semisupervised multiple instance learning , author=. AAAI , pages=

    work page

  51. [51]

    arXiv preprint arXiv:1607.03615 , year=

    Multiple-instance logistic regression with lasso penalty , author=. arXiv preprint arXiv:1607.03615 , year=

    work page arXiv

  52. [52]

    IEEE Transactions on Pattern Analysis and Machine Intelligence , volume=

    Fast multi-instance multi-label learning , author=. IEEE Transactions on Pattern Analysis and Machine Intelligence , volume=. 2018 , publisher=

    work page 2018

  53. [53]

    IEEE Transactions on Neural Networks and Learning Systems , volume=

    Scalable algorithms for multi-instance learning , author=. IEEE Transactions on Neural Networks and Learning Systems , volume=. 2016 , publisher=

    work page 2016

  54. [54]

    Information Sciences , volume=

    Bag similarity network for deep multi-instance learning , author=. Information Sciences , volume=. 2019 , publisher=

    work page 2019

  55. [55]

    IEEE Transactions on Knowledge and Data Engineering , year=

    Multiple-instance learning from unlabeled bags with pairwise similarity , author=. IEEE Transactions on Knowledge and Data Engineering , year=

    work page

  56. [56]

    Machine Learning , volume=

    Support-vector networks , author=. Machine Learning , volume=. 1995 , publisher=

    work page 1995

  57. [57]

    Department of Computer Science & Technology, Nanjing University, Tech

    Multi-instance learning: A survey , author=. Department of Computer Science & Technology, Nanjing University, Tech. Rep , volume=

    work page

  58. [58]

    Proceedings of the 25th ACM International on Conference on Information and Knowledge Management , pages=

    A multiple instance learning framework for identifying key sentences and detecting events , author=. Proceedings of the 25th ACM International on Conference on Information and Knowledge Management , pages=

    work page

  59. [59]

    Medical Image Analysis , volume=

    Not-so-supervised: a survey of semi-supervised, multi-instance, and transfer learning in medical image analysis , author=. Medical Image Analysis , volume=. 2019 , publisher=

    work page 2019

  60. [60]

    Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages=

    Chestx-ray8: Hospital-scale chest x-ray database and benchmarks on weakly-supervised classification and localization of common thorax diseases , author=. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages=

    work page

  61. [61]

    IEEE Transactions on Instrumentation and Measurement , volume=

    Deep multiple instance learning for automatic breast cancer assessment using digital mammography , author=. IEEE Transactions on Instrumentation and Measurement , volume=. 2022 , publisher=

    work page 2022

  62. [62]

    Annual Review of Statistics and Its Application , volume=

    Finite mixture models , author=. Annual Review of Statistics and Its Application , volume=. 2019 , publisher=

    work page 2019

  63. [63]

    Artificial Intelligence , volume=

    Multiple instance classification: Review, taxonomy and comparative study , author=. Artificial Intelligence , volume=. 2013 , publisher=

    work page 2013

  64. [64]

    R Journal , volume=

    milr: Multiple-instance logistic regression with lasso penalty , author=. R Journal , volume=

    work page

  65. [65]

    Neural Processing Letters , volume=

    Improve multi-instance neural networks through feature selection , author=. Neural Processing Letters , volume=. 2004 , publisher=

    work page 2004

  66. [66]

    Neural Processing Letters , volume=

    Adapting RBF neural networks to multi-instance learning , author=. Neural Processing Letters , volume=. 2006 , publisher=

    work page 2006

  67. [67]

    Pattern Recognition , volume=

    Revisiting multiple instance neural networks , author=. Pattern Recognition , volume=. 2018 , publisher=

    work page 2018

  68. [68]

    IEEE Transactions on Medical Imaging , volume=

    Constrained deep weak supervision for histopathology image segmentation , author=. IEEE Transactions on Medical Imaging , volume=. 2017 , publisher=

    work page 2017

  69. [69]

    Artificial Intelligence , volume=

    Solving the multiple instance problem with axis-parallel rectangles , author=. Artificial Intelligence , volume=. 1997 , publisher=

    work page 1997

  70. [70]

    Journal of Machine Learning Research , volume=

    Multiple-instance learning from distributions , author=. Journal of Machine Learning Research , volume=. 2016 , publisher=

    work page 2016

  71. [71]

    Proceedings of the 2020 4th International Conference on Computer Science and Artificial Intelligence , pages=

    A review of latest multi-instance learning , author=. Proceedings of the 2020 4th International Conference on Computer Science and Artificial Intelligence , pages=

    work page 2020

  72. [72]

    Nature Biomedical Engineering , volume=

    Fast and scalable search of whole-slide images via self-supervised deep learning , author=. Nature Biomedical Engineering , volume=. 2022 , publisher=

    work page 2022

  73. [73]

    Nature Machine Intelligence , volume=

    Pathologist-level interpretable whole-slide cancer diagnosis with deep learning , author=. Nature Machine Intelligence , volume=. 2019 , publisher=

    work page 2019

  74. [74]

    Nature Communications , volume=

    Human-interpretable image features derived from densely mapped cancer pathology slides predict diverse molecular phenotypes , author=. Nature Communications , volume=. 2021 , publisher=

    work page 2021

  75. [75]

    Nature Biomedical Engineering , volume=

    Data-efficient and weakly supervised computational pathology on whole-slide images , author=. Nature Biomedical Engineering , volume=. 2021 , publisher=

    work page 2021

  76. [76]

    IEEE Transactions on Medical Imaging , volume=

    Label cleaning multiple instance learning: Refining coarse annotations on single whole-slide images , author=. IEEE Transactions on Medical Imaging , volume=. 2022 , publisher=

    work page 2022

  77. [77]

    IJCAI , volume=

    AUC: a statistically consistent and more discriminating measure than accuracy , author=. IJCAI , volume=

    work page

  78. [78]

    arXiv preprint arXiv:2304.02269 , year=

    Distributed logistic regression for massive data with rare events , author=. arXiv preprint arXiv:2304.02269 , year=

    work page arXiv

  79. [79]

    International Conference on Machine Learning , pages=

    Logistic regression for massive data with rare events , author=. International Conference on Machine Learning , pages=. 2020 , organization=

    work page 2020

  80. [80]

    The Annals of Mathematical Statistics , volume=

    On estimation of a probability density function and mode , author=. The Annals of Mathematical Statistics , volume=. 1962 , publisher=

    work page 1962

Showing first 80 references.