pith. sign in

arxiv: 2607.02435 · v1 · pith:EASVNUTJnew · submitted 2026-07-02 · 💻 cs.CV · eess.IV

MARVEL: Margin-Aware Robust von Mises-Fischer Expert Learning for Long-Tailed Out-of-Distribution Detection

A.S. Anudeep , Vaanathi Sundaresan This is my paper

Pith reviewed 2026-07-03 14:57 UTC · model grok-4.3

classification 💻 cs.CV eess.IV
keywords out-of-distribution detectionlong-tailed learningmedical imagingvon Mises-Fisherexpert learningimbalanced classificationclinical decision support
0
0 comments X

The pith

A multi-expert framework with nonlinear von Mises-Fisher classifiers improves OOD detection on long-tailed medical datasets.

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

The paper presents MARVEL, a method for out-of-distribution detection in imbalanced medical imaging data. It relies on a nonlinear von Mises-Fisher classifier that learns non-linear boundaries and has an asymptotic link to cosine classifiers. A multi-expert setup lets margin-aware versions of this classifier specialize on different parts of the label distribution to address imbalance. An outlier expert is added to explicitly separate inliers from outliers. Results on RFMiD, ISIC2019, and NCTCRC datasets show reduced FPR95 rates, supporting safer clinical AI by identifying unfamiliar inputs.

Core claim

The framework of nonlinear von Mises-Fisher classifiers in a multi-expert margin-aware architecture combined with an outlier expert enables better out-of-distribution detection for long-tailed medical datasets, as evidenced by FPR95 reductions of 8.45%, 13.02%, and 36.90% on the three evaluated datasets.

What carries the argument

Nonlinear von Mises-Fisher (NvMF) classifier that learns non-linear decision boundaries with asymptotic connection to cosine classifiers, used in a multi-expert framework with margin awareness and an outlier expert.

If this is right

  • Each component of the framework contributes to improved OOD detection as shown by ablations.
  • The approach handles label imbalance by specializing experts on different distribution regions.
  • Explicit outlier training strengthens distinction between in-distribution and out-of-distribution data.
  • Results hold across multiple clinically relevant medical datasets with diverse OOD scenarios.

Where Pith is reading between the lines

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

  • The method may generalize to other domains with long-tailed distributions beyond medical imaging.
  • Simplifying the nonlinear aspect using the asymptotic connection could reduce computational overhead in deployment.
  • Further tests with real-world clinical workflows could reveal additional benefits for deferral systems.

Load-bearing premise

The theoretical asymptotic connection of the nonlinear von Mises-Fisher classifier to cosine classifiers enables effective non-linear boundary learning in practice for out-of-distribution detection.

What would settle it

Observing no improvement in FPR95 or failure of the NvMF component to learn useful non-linear boundaries on the RFMiD, ISIC2019, or NCTCRC datasets would falsify the claim.

Figures

Figures reproduced from arXiv: 2607.02435 by A.S. Anudeep, Vaanathi Sundaresan.

Figure 1
Figure 1. Figure 1: Overview of the proposed long-tailed OOD detection framework. (a) Schematic of the possible [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Dataset composition and evaluation setup for long-tailed OOD detection. (a) Shows represen [PITH_FULL_IMAGE:figures/full_fig_p012_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Head, mid, tail class accuracy comparison [PITH_FULL_IMAGE:figures/full_fig_p016_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: shows the impact of number of experts within the ensemble on ID classification and OOD detection across the three datasets. Increasing the number of experts yields consistent perfor￾mance gains up to 3 experts. For ID classifica￾tion (left), moving from a single expert to three experts improves accuracy on RFMiD (≈65% → 67.5%), ISIC2019 (≈67.5% → 73%), and NCTCRC (≈73% → 77%). Similar trends are observed f… view at source ↗
Figure 6
Figure 6. Figure 6: Risk-coverage curves for OOD detection across RFMiD, ISIC2019, and NCTCRC datasets. The x-axis represents coverage and the y-axis de￾notes risk and the individual legends show the AURC (area under the risk-coverage curve) values in the corresponding plot colours. Across datasets, RFMiD demonstrates the most favorable behaviour. FarOOD, NearOOD1, and NearOOD2 curves largely overlap and maintain near-zero ri… view at source ↗
Figure 7
Figure 7. Figure 7: t-SNE visualization of learned feature representations in the NCTCRC dataset. Representation [PITH_FULL_IMAGE:figures/full_fig_p020_7.png] view at source ↗
read the original abstract

For clinical deployment, it is essential that automated diagnostic systems remain reliable when confronted with previously unseen cases, yet deep models routinely misclassify out-of-distribution (OOD) inputs with high confidence, underscoring the need for more robust OOD detection methods. Although substantial effort has been devoted to improving model robustness, most of the existing literature assumes balanced datasets, evaluates OOD detection on coarse or non-clinical OOD sources, or lacks comprehensive assessment across diverse OOD scenarios. To address the gaps, we propose a novel methodology trained on diverse and imbalanced medical datasets and evaluated across a clinically reflective OOD spectrum. Our framework comprises three key components: (1) a Nonlinear von Mises-Fisher (NvMF) classifier capable of learning non-linear decision boundaries, with theoretical proof of its asymptotic connection to cosine classifiers; (2) a multi-expert framework in which margin-aware NvMF classifiers specialise in different regions of label distribution to better handle imbalance; and (3) an outlier expert trained explicitly to distinguish inlier from outlier data, thereby strengthening OOD detection. Evaluation on RFMiD, ISIC2019, and NCTCRC datasets demonstrates consistent improvements over state-of-the-art methods, achieving mean FPR95 reductions of 8.45%, 13.02%, and 36.90% respectively. These gains are further supported by comprehensive ablations that validated the contributions of each component. This enables reliable identification of unfamiliar cases for deferral to clinicians, supporting safer AI-assisted diagnosis in real-world workflows. Our code is available at https://github.com/redboxup/MARVEL.

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

3 major / 2 minor

Summary. The paper proposes MARVEL, a framework for long-tailed OOD detection on imbalanced medical imaging datasets. It consists of (1) a Nonlinear von Mises-Fisher (NvMF) classifier with a theoretical proof of asymptotic connection to cosine classifiers for learning non-linear boundaries, (2) a multi-expert system of margin-aware NvMF classifiers specialized to different regions of the label distribution, and (3) an outlier expert trained to distinguish inliers from outliers. Experiments on RFMiD, ISIC2019, and NCTCRC report mean FPR95 reductions of 8.45%, 13.02%, and 36.90% over SOTA, supported by ablations; code is released.

Significance. If the empirical gains and theoretical connection hold under rigorous scrutiny, the work would advance reliable OOD detection for clinical long-tailed settings, directly supporting safer deferral workflows. Releasing code strengthens the contribution.

major comments (3)
  1. [Abstract / §3] Abstract and §3 (theoretical component): the claimed asymptotic connection of NvMF to cosine classifiers is presented as enabling non-linear boundaries, yet no equation or proof sketch is supplied in the summary to verify whether the connection is parameter-free or holds under the nonlinear extension; this is load-bearing for component (1).
  2. [§4] §4 (experiments): the reported FPR95 reductions lack any description of baselines, training protocols, statistical significance, or OOD source construction, preventing assessment of whether the multi-expert and outlier-expert gains are robust or dataset-specific.
  3. [§3.3] §3.3 (outlier expert): the explicit training of a dedicated outlier expert is central to the OOD claim, but no loss function, sampling strategy, or integration details are given, raising the risk that performance improvements are driven by this ad-hoc component rather than the NvMF core.
minor comments (2)
  1. Define NvMF and all acronyms at first use; ensure figure captions fully describe axes and legend entries for the ablation studies.
  2. [Abstract] The abstract states 'comprehensive ablations' but does not indicate which tables or figures contain them; cross-reference explicitly.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive feedback. Below we respond point-by-point to the major comments.

read point-by-point responses

  1. Referee: [Abstract / §3] Abstract and §3 (theoretical component): the claimed asymptotic connection of NvMF to cosine classifiers is presented as enabling non-linear boundaries, yet no equation or proof sketch is supplied in the summary to verify whether the connection is parameter-free or holds under the nonlinear extension; this is load-bearing for component (1).

    Authors: Section 3 contains the full proof establishing the asymptotic connection to cosine classifiers, demonstrating that the link is parameter-free and extends to the nonlinear case. The abstract summarizes this result at a high level. We will add a concise equation and one-sentence proof sketch to the abstract to make the theoretical claim self-contained. revision: partial

  2. Referee: [§4] §4 (experiments): the reported FPR95 reductions lack any description of baselines, training protocols, statistical significance, or OOD source construction, preventing assessment of whether the multi-expert and outlier-expert gains are robust or dataset-specific.

    Authors: Section 4 and the supplementary material specify the SOTA baselines, training protocols with hyperparameters, statistical significance via repeated runs with standard deviations, and the construction of clinically relevant OOD sources. We will expand the experimental section with additional explicit details on these elements to improve clarity and reproducibility. revision: yes

  3. Referee: [§3.3] §3.3 (outlier expert): the explicit training of a dedicated outlier expert is central to the OOD claim, but no loss function, sampling strategy, or integration details are given, raising the risk that performance improvements are driven by this ad-hoc component rather than the NvMF core.

    Authors: Section 3.3 defines the outlier expert's loss, sampling procedure, and integration via the multi-expert gating mechanism. To eliminate any perception of ad-hoc design, we will insert the explicit loss equation and a schematic of the integration in the revised version. revision: yes

Circularity Check

0 steps flagged

No circularity in derivation chain from provided text

full rationale

The abstract and description present three components (NvMF classifier with claimed theoretical asymptotic connection to cosine classifiers, multi-expert margin-aware setup, and outlier expert) along with dataset evaluations showing FPR95 improvements. No equations, self-citations, fitted parameters presented as predictions, or self-definitional steps are quoted or visible in the given material. The theoretical proof is asserted as independent support rather than a reduction to inputs, and external benchmarks (RFMiD, ISIC2019, NCTCRC) are referenced, making the central claims self-contained against the provided excerpts.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

Limited information from abstract; the outlier expert is presented as a novel component without independent evidence provided.

axioms (1)
  • domain assumption The NvMF classifier has an asymptotic connection to cosine classifiers
    Mentioned as having theoretical proof in the abstract.
invented entities (1)
  • Outlier expert no independent evidence
    purpose: To distinguish inlier from outlier data
    Introduced as a new component in the framework.

pith-pipeline@v0.9.1-grok · 5834 in / 1401 out tokens · 49681 ms · 2026-07-03T14:57:12.447505+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

80 extracted references · 22 canonical work pages · 3 internal anchors

  1. [1]

    Barry C. Arnold. Pareto Distributions. 2nd ed. Chapman and Hall/CRC, 2015.doi:10.1201/ b18141

    2015

  2. [2]

    Christopher M. Bishop. Pattern Recognition and Machine Learning (Information Science and Statistics). Berlin, Heidelberg: Springer-Verlag, 2006.isbn: 0387310738

    2006

  3. [3]

    Albumentations: fast and flexible image augmentations

    A. Buslaev et al. “Albumentations: fast and flexible image augmentations”. In: arXiv preprint arXiv:1809.06839 (2018). eprint:1809.06839

    work page arXiv 2018

  4. [4]

    Learning Imbalanced Datasets with Label-Distribution-Aware Margin Loss

    Kaidi Cao et al. “Learning Imbalanced Datasets with Label-Distribution-Aware Margin Loss”. In: Advances in Neural Information Processing Systems. 2019

    2019

  5. [5]

    Dataset of fundus images for the diagnosis of ocular toxoplasmosis

    Olivia Cardozo et al. “Dataset of fundus images for the diagnosis of ocular toxoplasmosis”. In: Data in Brief 48 (2023), p. 109056.issn: 2352-3409

    2023

  6. [6]

    The DFUC 2020 Dataset: Analysis Towards Diabetic Foot Ulcer Detection

    B. Cassidy et al. “The DFUC 2020 Dataset: Analysis Towards Diabetic Foot Ulcer Detection”. In: touchREV Endocrinology 17.1 (Apr. 2021), pp. 5–11.doi:10.17925/EE.2021.17.1.5

    work page doi:10.17925/ee.2021.17.1.5 2020

  7. [7]

    N. C. F. Codella et al. Skin Lesion Analysis Toward Melanoma Detection: A Challenge at the 2017 International Symposium on Biomedical Imaging (ISBI), Hosted by the International Skin Imaging Collaboration (ISIC). arXiv:1710.05006. 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  8. [8]

    iNaturalist Research-grade Observations

    iNaturalist contributors. iNaturalist Research-grade Observations. 2024

    2024

  9. [9]

    Class-Balanced Loss Based on Effective Number of Samples

    Yin Cui et al. “Class-Balanced Loss Based on Effective Number of Samples”. In: CVPR. 2019

    2019

  10. [10]

    Extremely Simple Activation Shaping for Out-of-Distribution Detection

    Andrija Djurisic et al. “Extremely Simple Activation Shaping for Out-of-Distribution Detection”. In: The Eleventh International Conference on Learning Representations. 2023.url:https://openreview. net/forum?id=ndYXTEL6cZz

    2023

  11. [11]

    VOS: Learning What You Don’t Know by Virtual Outlier Synthesis

    Xuefeng Du et al. “VOS: Learning What You Don’t Know by Virtual Outlier Synthesis”. In: Proceedings of the International Conference on Learning Representations (2022)

    2022

  12. [12]

    Dermatologist-level classification of skin cancer with deep neural networks

    Andre Esteva et al. “Dermatologist-level classification of skin cancer with deep neural networks”. In: Nature 542.7639 (2017), pp. 115–118.doi:10.1038/nature21056

    work page doi:10.1038/nature21056 2017

  13. [13]

    Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning

    Yarin Gal and Zoubin Ghahramani. “Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning”. In:Proceedings of The 33rd International Conference on Machine Learning. Ed. by Maria Florina Balcan and Kilian Q. Weinberger. Vol. 48. Proceedings of Machine Learning Research. New York, New York, USA: PMLR, 20–22 Jun 2016, pp. 1050–1059

    2016

  14. [14]

    The generalized von Mises distribution

    Riccardo Gatto and Sreenivasa Rao Jammalamadaka. “The generalized von Mises distribution”. In: Statistical Methodology 4.3 (2007), pp. 341–353

    2007

  15. [15]

    Dynamic Few-Shot Visual Learning Without Forgetting

    Spyros Gidaris and Nikos Komodakis. “Dynamic Few-Shot Visual Learning Without Forgetting”. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). June 2018

    2018

  16. [16]

    Does your dermatology classifier know what it doesn’t know? Detect- ing the long-tail of unseen conditions

    Abhijit Guha Roy et al. “Does your dermatology classifier know what it doesn’t know? Detect- ing the long-tail of unseen conditions”. In: Medical Image Analysis 75 (2022), p. 102274.issn: 1361-8415.doi:https : / / doi . org / 10 . 1016 / j . media . 2021 . 102274.url:https : / / www . sciencedirect.com/science/article/pii/S1361841521003194

    2022

  17. [17]

    Long-Tailed Out-of-Distribution Detection: Prioritizing Attention to Tail

    Yina He et al. “Long-Tailed Out-of-Distribution Detection: Prioritizing Attention to Tail”. In: Proceedings of the AAAI Conference on Artificial Intelligence 39.3 (Apr. 2025), pp. 3446–3454.doi: 10.1609/aaai.v39i3.32357.url:https://ojs.aaai.org/index.php/AAAI/article/view/ 32357

    work page doi:10.1609/aaai.v39i3.32357.url:https://ojs.aaai.org/index.php/aaai/article/view/ 2025

  18. [18]

    A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks

    Dan Hendrycks and Kevin Gimpel. “A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks”. In:Proceedings of International Conference on Learning Representations (2017)

    2017

  19. [19]

    Deep Anomaly Detection with Outlier Exposure

    Dan Hendrycks, Mantas Mazeika, and Thomas Dietterich. “Deep Anomaly Detection with Outlier Exposure”. In: Proceedings of the International Conference on Learning Representations (2019)

    2019

  20. [20]

    BCN20000: Dermoscopic lesions in the wild

    C. Hern´ andez-P´ erez et al. “BCN20000: Dermoscopic lesions in the wild”. In: Scientific Data 11.1 (June 2024), p. 641.doi:10.1038/s41597-024-01921-3

    work page doi:10.1038/s41597-024-01921-3 2024

  21. [21]

    Generalized ODIN: Detecting Out-of-Distribution Image Without Learning From Out-of-Distribution Data

    Yen-Chang Hsu et al. “Generalized ODIN: Detecting Out-of-Distribution Image Without Learning From Out-of-Distribution Data”. In:Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). June 2020. 22

    2020

  22. [22]

    J. N. Kather, N. Halama, and A. Marx. 100,000 histological images of human colorectal cancer and healthy tissue. Data set. Version v0.1. 2018

    2018

  23. [23]

    Kather, Nikolaus Halama, and Andreas Marx

    Jakob N. Kather, Nikolaus Halama, and Andreas Marx. 100,000 histological images of human colorectal cancer and healthy tissue (v0.1) [Data set]. 2018.doi:10.5281/zenodo.1214456.url:https://doi.org/10.5281/zenodo.1214456

    work page doi:10.5281/zenodo.1214456.url:https://doi.org/10.5281/zenodo.1214456 2018

  24. [24]

    Identifying Medical Diagnoses and Treatable Diseases by Image-Based Deep Learning

    Daniel S. Kermany et al. “Identifying Medical Diagnoses and Treatable Diseases by Image-Based Deep Learning”. In: Cell 172.5 (2018), 1122–1131.e9.issn: 0092-8674.doi:10 . 1016 / j . cell . 2018.02.010

    2018

  25. [25]

    Supervised contrastive learning,

    Prannay Khosla et al. “Supervised Contrastive Learning”. In: arXiv preprint arXiv:2004.11362 (2020)

    work page arXiv 2004

  26. [26]

    Adam: A Method for Stochastic Optimization

    Diederik P. Kingma and Jimmy Lei Ba. “Adam: A Method for Stochastic Optimization”. In: International Conference on Learning Representations (ICLR). Ithaca, NY: ArXiv, 2015

    2015

  27. [27]

    Generalized Outlier Exposure: Towards a trust- worthy out-of-distribution detector without sacrificing accuracy

    Jiin Koo, Sungjoon Choi, and Sangheum Hwang. “Generalized Outlier Exposure: Towards a trust- worthy out-of-distribution detector without sacrificing accuracy”. In: Neurocomputing 577 (2024), p. 127371.issn: 0925-2312

    2024

  28. [28]

    Learning multiple layers of features from tiny images

    Alex Krizhevsky. Learning multiple layers of features from tiny images. Technical Report. Univer- sity of Toronto, 2009

    2009

  29. [29]

    ImageNet Classification with Deep Con- volutional Neural Networks

    Alex Krizhevsky, Ilya Sutskever, and Geoffrey E Hinton. “ImageNet Classification with Deep Con- volutional Neural Networks”. In: Advances in Neural Information Processing Systems. Ed. by F. Pereira et al. Vol. 25. Curran Associates, Inc., 2012.url:https://proceedings.neurips.cc/ paper_files/paper/2012/file/c399862d3b9d6b76c8436e924a68c45b-Paper.pdf

    2012

  30. [30]

    Simple and Scalable Predic- tive Uncertainty Estimation using Deep Ensembles

    Balaji Lakshminarayanan, Alexander Pritzel, and Charles Blundell. “Simple and Scalable Predic- tive Uncertainty Estimation using Deep Ensembles”. In:Advances in Neural Information Processing Systems. Ed. by I. Guyon et al. Vol. 30. Curran Associates, Inc., 2017

    2017

  31. [31]

    A Simple Unified Framework for Detecting Out-of-Distribution Samples and Adversarial Attacks

    Kimin Lee et al. “A Simple Unified Framework for Detecting Out-of-Distribution Samples and Adversarial Attacks”. In: Advances in Neural Information Processing Systems. Ed. by S. Bengio et al. Vol. 31. Curran Associates, Inc., 2018

    2018

  32. [32]

    Enhancing The Reliability of Out-of-distribution Image Detection in Neural Networks

    Shiyu Liang, Yixuan Li, and R. Srikant. “Enhancing The Reliability of Out-of-distribution Image Detection in Neural Networks”. In: International Conference on Learning Representations. 2018. url:https://openreview.net/forum?id=H1VGkIxRZ

    2018

  33. [33]

    Microsoft COCO: Common Objects in Context

    Tsung-Yi Lin et al. “Microsoft COCO: Common Objects in Context”. In: European Conference on Computer Vision (ECCV). 2014, pp. 740–755.doi:10.1007/978-3-319-10602-1_48

    work page doi:10.1007/978-3-319-10602-1_48 2014

  34. [34]

    Microsoft coco: Common objects in context

    Tsung-Yi Lin et al. “Microsoft coco: Common objects in context”. In: European conference on computer vision. Springer. 2014, pp. 740–755

    2014

  35. [35]

    A survey on deep learning in medical image analysis

    Geert Litjens et al. “A survey on deep learning in medical image analysis”. In: Medical Image Analysis 42 (2017), pp. 60–88.issn: 1361-8415.doi:https://doi.org/10.1016/j.media.2017.07.005. url:https://www.sciencedirect.com/science/article/pii/S1361841517301135

    work page doi:10.1016/j.media.2017.07.005 2017

  36. [36]

    DeepDRiD: Diabetic Retinopathy—Grading and Image Quality Estimation Challenge

    Ruhan Liu et al. “DeepDRiD: Diabetic Retinopathy—Grading and Image Quality Estimation Challenge”. In: Patterns 3.6 (2022), p. 100512.issn: 2666-3899.doi:https : / / doi . org / 10 . 1016/j.patter.2022.100512.url:https://www.sciencedirect.com/science/article/pii/ S2666389922001040

    work page arXiv 2022

  37. [37]

    Energy-based Out-of-distribution Detection

    Weitang Liu et al. “Energy-based Out-of-distribution Detection”. In: Advances in Neural Information Processing Systems (2020)

    2020

  38. [38]

    SphereFace: Deep Hypersphere Embedding for Face Recognition

    Weiyang Liu et al. “SphereFace: Deep Hypersphere Embedding for Face Recognition”. In:The IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2017

    2017

  39. [39]

    A comparison of deep learning performance against health-care profes- sionals in detecting diseases from medical imaging: a systematic review and meta-analysis

    Xiaoxuan Liu et al. “A comparison of deep learning performance against health-care profes- sionals in detecting diseases from medical imaging: a systematic review and meta-analysis”. In: The Lancet Digital Health 1.6 (2019), e271–e297.doi:10.1016/S2589-7500(19)30123-2

    work page doi:10.1016/s2589-7500(19)30123-2 2019

  40. [40]

    Large-Scale Long-Tailed Recognition in an Open World

    Ziwei Liu et al. “Large-Scale Long-Tailed Recognition in an Open World”. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2019

    2019

  41. [41]

    Learning with Mixture of Prototypes for Out-of-Distribution Detection

    Haodong Lu et al. “Learning with Mixture of Prototypes for Out-of-Distribution Detection”. In: The Twelfth International Conference on Learning Representations. 2024.url:https://openreview. net/forum?id=uNkKaD3MCs. 23

    2024

  42. [43]

    Out-of-Distribution Detection for Long-Tailed and Fine-Grained Skin Lesion Images

    Deval Mehta et al. “Out-of-Distribution Detection for Long-Tailed and Fine-Grained Skin Lesion Images”. In: Medical Image Computing and Computer Assisted Intervention – MICCAI 2022. Ed. by Linwei Wang et al. Cham: Springer Nature Switzerland, 2022, pp. 732–742.isbn: 978-3-031- 16431-6

    2022

  43. [44]

    Long-tailed Visual Recognition via Gaussian Clouded Logit Adjustment

    Yang Lu Mengke Li Yiu-ming Cheung. “Long-tailed Visual Recognition via Gaussian Clouded Logit Adjustment”. In: CVPR. 2022, pp. 6929–6938

    2022

  44. [45]

    Long-tail learning via logit adjustment

    Aditya Krishna Menon et al. “Long-tail learning via logit adjustment”. In: International Conference on Learning Representations. 2021.url:https://openreview.net/forum?id=37nvvqkCo5

    2021

  45. [46]

    Out-of-distribution detection in long-tailed recognition with calibrated outlier class learning

    Wenjun Miao et al. “Out-of-distribution detection in long-tailed recognition with calibrated outlier class learning”. In: Proceedings of the AAAI Conference on Artificial Intelligence. Vol. 38. 2024

    2024

  46. [47]

    POEM: Out-of-Distribution Detection with Posterior Sam- pling

    Yifei Ming, Ying Fan, and Yixuan Li. “POEM: Out-of-Distribution Detection with Posterior Sam- pling”. In: Proceedings of the 39th International Conference on Machine Learning. Ed. by Kama- lika Chaudhuri et al. Vol. 162. Proceedings of Machine Learning Research. PMLR, 17–23 Jul 2022, pp. 15650–15665.url:https://proceedings.mlr.press/v162/ming22a.html

    2022

  47. [48]

    How to Exploit Hyperspherical Embeddings for Out-of-Distribution Detection?

    Yifei Ming et al. “How to Exploit Hyperspherical Embeddings for Out-of-Distribution Detection?” In: The Eleventh International Conference on Learning Representations. 2023

    2023

  48. [49]

    Reading Digits in Natural Images with Unsupervised Feature Learning

    Yuval Netzer et al. Reading Digits in Natural Images with Unsupervised Feature Learning. 2011

    2011

  49. [50]

    VinDr-CXR: An open dataset of chest X-rays with radiologist’s annotations

    H. Q. Nguyen, K. Lam, L. T. Le, et al. “VinDr-CXR: An open dataset of chest X-rays with radiologist’s annotations”. In: Scientific Data 9 (2022), p. 429

    2022

  50. [51]

    Chapter 10 Bessel Functions

    NIST Digital Library of Mathematical Functions. Chapter 10 Bessel Functions. Revision 1.1.5. 2020.url:https://dlmf.nist.gov/10.40(visited on 10/27/2023)

    2020

  51. [52]

    Retinal Fundus Multi-disease Image Dataset (RFMiD)

    Samiksha Pachade et al. Retinal Fundus Multi-disease Image Dataset (RFMiD). 2020

    2020

  52. [53]

    Retinal Fundus Multi-disease Image Dataset (RFMiD)

    Samiksha Pachade et al. Retinal Fundus Multi-disease Image Dataset (RFMiD). 2020.doi:10 . 21227/s3g7-st65.url:https://dx.doi.org/10.21227/s3g7-st65

    work page doi:10.21227/s3g7-st65 2020

  53. [54]

    PAD-UFES-20: A skin lesion dataset composed of patient data and clinical images collected from smartphones

    Andre G.C. Pacheco et al. “PAD-UFES-20: A skin lesion dataset composed of patient data and clinical images collected from smartphones”. In: Data in Brief 32 (2020), p. 106221.issn: 2352- 3409.doi:https://doi.org/10.1016/j.dib.2020.106221.url:https://www.sciencedirect. com/science/article/pii/S235234092031115X

    work page doi:10.1016/j.dib.2020.106221.url:https://www.sciencedirect 2020

  54. [55]

    Nearest Neighbor Guidance for Out-of- Distribution Detection

    Jaewoo Park, Yoon Gyo Jung, and Andrew Beng Jin Teoh. “Nearest Neighbor Guidance for Out-of- Distribution Detection”. In:Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). Oct. 2023, pp. 1686–1695

    2023

  55. [56]

    Hady Ahmady Phoulady and Peter R. Mouton. A New Cervical Cytology Dataset for Nucleus Detection and Image Classification (Cervix93) and Methods for Cervical Nucleus Detection

  56. [57]

    arXiv:1811.09651 [cs.CV].url:https://arxiv.org/abs/1811.09651

    work page internal anchor Pith review Pith/arXiv arXiv

  57. [58]

    ImageNet Large Scale Visual Recognition Challenge

    Olga Russakovsky et al. “ImageNet Large Scale Visual Recognition Challenge”. In: International Journal of Computer Vision (IJCV) 115.3 (2015), pp. 211–252

    2015

  58. [59]

    SSD: A Unified Framework for Self-Supervised Outlier Detection

    Vikash Sehwag, Mung Chiang, and Prateek Mittal. “SSD: A Unified Framework for Self-Supervised Outlier Detection”. In: International Conference on Learning Representations. 2021.url:https: //openreview.net/forum?id=v5gjXpmR8J

    2021

  59. [60]

    EBHI-Seg: A novel enteroscope biopsy histopathological hematoxylin and eosin image dataset for image segmentation tasks

    L. Shi et al. “EBHI-Seg: A novel enteroscope biopsy histopathological hematoxylin and eosin image dataset for image segmentation tasks”. In: Frontiers in Medicine 10 (2023), p. 1114673

    2023

  60. [61]

    ReAct: Out-of-distribution Detection With Rectified Activations

    Yiyou Sun, Chuan Guo, and Yixuan Li. “ReAct: Out-of-distribution Detection With Rectified Activations”. In: Advances in Neural Information Processing Systems. 2021

    2021

  61. [62]

    DICE: Leveraging Sparsification for Out-of-Distribution Detection

    Yiyou Sun and Yixuan Li. “DICE: Leveraging Sparsification for Out-of-Distribution Detection”. In: European Conference on Computer Vision. 2022

    2022

  62. [63]

    Out-of-distribution Detection with Deep Nearest Neighbors

    Yiyou Sun et al. “Out-of-distribution Detection with Deep Nearest Neighbors”. In: ICML (2022)

    2022

  63. [64]

    Non-parametric Outlier Synthesis

    Leitian Tao et al. “Non-parametric Outlier Synthesis”. In: The Eleventh International Conference on Learning Representations. 2023.url:https://openreview.net/forum?id=JHklpEZqduQ. 24

    2023

  64. [65]

    The HAM10000 dataset, a large collection of multi- source dermatoscopic images of common pigmented skin lesions

    P. Tschandl, C. Rosendahl, and H. Kittler. “The HAM10000 dataset, a large collection of multi- source dermatoscopic images of common pigmented skin lesions”. In: Scientific Data 5 (2018), p. 180161.doi:10.1038/sdata.2018.161

    work page doi:10.1038/sdata.2018.161 2018

  65. [66]

    Open-Set Recognition: a Good Closed-Set Classifier is All You Need?

    Sagar Vaze et al. “Open-Set Recognition: a Good Closed-Set Classifier is All You Need?” In: International Conference on Learning Representations. 2022

    2022

  66. [67]

    ViM: Out-Of-Distribution with Virtual-logit Matching

    Haoqi Wang et al. “ViM: Out-Of-Distribution with Virtual-logit Matching”. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2022

    2022

  67. [68]

    Partial and Asymmetric Contrastive Learning for Out-of-Distribution Detec- tion in Long-Tailed Recognition

    Haotao Wang et al. “Partial and Asymmetric Contrastive Learning for Out-of-Distribution Detec- tion in Long-Tailed Recognition”. In:International Conference on Machine Learning. 2022, pp. 23446– 23458

    2022

  68. [69]

    MIHIC: a multiplex IHC histopathological image classification dataset for lung cancer immune microenvironment quantification

    R. Wang et al. “MIHIC: a multiplex IHC histopathological image classification dataset for lung cancer immune microenvironment quantification”. In: Frontiers in Immunology 15 (Feb. 2024), p. 1334348.doi:10.3389/fimmu.2024.1334348

    work page doi:10.3389/fimmu.2024.1334348 2024

  69. [70]

    Understanding Contrastive Representation Learning through Alignment and Uniformity on the Hypersphere

    Tongzhou Wang and Phillip Isola. “Understanding Contrastive Representation Learning through Alignment and Uniformity on the Hypersphere”. In:International Conference on Machine Learning. PMLR. 2020, pp. 9929–9939

    2020

  70. [71]

    ChestX-Ray8: Hospital-Scale Chest X-Ray Database and Benchmarks on Weakly-Supervised Classification and Localization of Common Thorax Diseases

    Xiaosong Wang et al. “ChestX-Ray8: Hospital-Scale Chest X-Ray Database and Benchmarks on Weakly-Supervised Classification and Localization of Common Thorax Diseases”. In:2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2017, pp. 3462–3471.doi:10.1109/CVPR.2017.369

    work page doi:10.1109/cvpr.2017.369 2017

  71. [72]

    EAT: Towards Long-Tailed Out-of-Distribution Detection

    Tong Wei, Bo-Lin Wang, and Min-Ling Zhang. “EAT: Towards Long-Tailed Out-of-Distribution Detection”. In: Proceedings of the 38th AAAI Conference on Artificial Intelligence (2024)

    2024

  72. [73]

    Out-of-distribution detection with in-distribution voting using the med- ical example of chest x-ray classification

    Alessandro Wollek et al. “Out-of-distribution detection with in-distribution voting using the med- ical example of chest x-ray classification”. In: Medical Physics 51.4 (2024), pp. 2721–2732.doi: https://doi.org/10.1002/mp.16790. eprint:https://aapm.onlinelibrary.wiley.com/doi/ pdf/10.1002/mp.16790.url:https://aapm.onlinelibrary.wiley.com/doi/abs/10.1002/...

    work page doi:10.1002/mp.16790 2024

  73. [74]

    Out-of-distribution detection with in-distribution voting using the medical example of chest x-ray classification

    Anne Wollek et al. “Out-of-distribution detection with in-distribution voting using the medical example of chest x-ray classification”. In: Medical Physics 51 (2024), pp. 2721–2732.doi:10 . 1002/mp.16790

    2024

  74. [75]

    Scaling for Training Time and Post-hoc Out-of-distribution Detection Enhance- ment

    Kai Xu et al. “Scaling for Training Time and Post-hoc Out-of-distribution Detection Enhance- ment”. In: The Twelfth International Conference on Learning Representations. 2024.url:https: //openreview.net/forum?id=RDSTjtnqCg

    2024

  75. [76]

    Generalized Out-of-Distribution Detection: A Survey

    Jingkang Yang et al. Generalized Out-of-Distribution Detection: A Survey. 2024. arXiv:2110 . 11334 [cs.CV]

    2024

  76. [77]

    Semantically Coherent Out-of-Distribution Detection

    Jingkang Yang et al. “Semantically Coherent Out-of-Distribution Detection”. In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). Oct. 2021, pp. 8301–8309

    2021

  77. [78]

    LSUN: Construction of a Large-scale Image Dataset using Deep Learning with Humans in the Loop

    Fisher Yu et al. “LSUN: Construction of a Large-scale Image Dataset using Deep Learning with Humans in the Loop”. In: arXiv preprint arXiv:1506.03365 (2015)

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  78. [79]

    Unsupervised Out-of-Distribution Detection by Maximum Clas- sifier Discrepancy

    Qing Yu and Kiyoharu Aizawa. “ Unsupervised Out-of-Distribution Detection by Maximum Clas- sifier Discrepancy ”. In: 2019 IEEE/CVF International Conference on Computer Vision (ICCV). Los Alamitos, CA, USA: IEEE Computer Society, Nov. 2019, pp. 9517–9525.doi:10.1109/ICCV. 2019.00961.url:https://doi.ieeecomputersociety.org/10.1109/ICCV.2019.00961

    work page doi:10.1109/iccv 2019

  79. [80]

    Mixture Outlier Exposure: Towards Out-of-Distribution Detection in Fine- Grained Environments

    Jingyang Zhang et al. “Mixture Outlier Exposure: Towards Out-of-Distribution Detection in Fine- Grained Environments”. In:Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV). Jan. 2023, pp. 5531–5540

    2023

  80. [81]

    A Review of Deep Learning in Medical Imaging: Imaging Traits, Technology Trends, Case Studies With Progress Highlights, and Future Promises

    S. Kevin Zhou et al. “A Review of Deep Learning in Medical Imaging: Imaging Traits, Technology Trends, Case Studies With Progress Highlights, and Future Promises”. In:Proceedings of the IEEE 109.5 (2021), pp. 820–838.doi:10.1109/JPROC.2021.3054390. 25

    work page doi:10.1109/jproc.2021.3054390 2021