Pith. sign in

REVIEW 3 major objections 4 minor 47 references

Geometry-Aware Bayesian Quantification via Compositional Data Analysis

T0 review · 3 major / 4 minor · reviewed 2026-07-11 · grok-4.5

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

Pith's one-line read Classifier posteriors live on the simplex; a shrinkage-regularized log-ratio KDE gives geometry-aware multiclass prevalence estimates and Bayesian posteriors under label shift that often beat Euclidean Gaussian KDE.

desk verdict Solid, usable upgrade to KDEy: shrinkage-regularized Aitchison KDE plus the first practical Bayesian continuous multivariate quantifier, with clean math and thorough multiclass evidence. read the letter →

arxiv 2607.04977 v1 pith:J632IS7T submitted 2026-07-06 cs.LG stat.ML

Alejandro Moreo , Pablo González , Juan José del Coz This is my paper
classification cs.LGstat.ML
keywords quantificationlabelshiftclassprevalenceestimationkerneldensitycompositionaldataanalysisAitchisongeometryBayesianposteriorcalibration
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

When a model’s class distribution changes between training and deployment (label shift), the first practical step is to estimate the new class prevalences from unlabeled target data. Continuous KDE methods that model the density of multiclass classifier posteriors have become strong tools for this, but they usually place ordinary Gaussian kernels in Euclidean space. Those kernels ignore that posteriors are compositional vectors on the probability simplex and leak mass outside its boundaries, especially near vertices where confident predictions concentrate. This paper replaces that geometry with Aitchison log-ratio coordinates plus a shrinkage map that pulls compositions toward the barycenter, then derives both maximum-likelihood point estimates and tempered Bayesian posteriors over prevalences. Across 42 tabular, text, and image datasets the resulting estimator is competitive with state-of-the-art quantifiers, frequently improves on standard KDE baselines, and ranks among the strongest Bayesian methods while respecting the true support of the data.

What carries the argument

Shrinkage-regularized geometry-aware KDE: each posterior p is first shrunk by T_λ(p)=(1-λ)p+λu toward the barycenter, mapped by the centered log-ratio transform into Euclidean coordinates, then smoothed by a Gaussian kernel whose bandwidth is rescaled as h_eff=(1-λ)h. The resulting class-conditional densities form a mixture whose likelihood (Jacobian-free for inference over π) supports both point estimation and Bayesian posterior sampling.

What would settle it

On a suite of datasets deliberately constructed with substantial covariate shift or sparse joint shift (where class-conditional posteriors are no longer domain-invariant), check whether the geometry-aware KDE still improves over Euclidean KDEy and EMQ; systematic underperformance would falsify the claim that the simplex-aware construction is the decisive advantage.

Watch

Extended reading notes

Core claim

Modeling multiclass classifier posteriors with a shrinkage-regularized log-ratio (Aitchison) kernel density estimator yields a mixture model whose maximum-likelihood and Bayesian solutions for class prevalences are competitive with, and often superior to, Euclidean Gaussian KDE and other standard quantifiers under label shift, while correctly staying inside the simplex.

Load-bearing premise

Class-conditional distributions of classifier posteriors stay the same across domains, so all quantification can be performed entirely in posterior space under pure label shift.

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

3 major / 4 minor

Summary. The paper addresses multiclass quantification (class prevalence estimation) under label shift by modeling classifier posterior vectors as compositional data on the simplex. It replaces Euclidean Gaussian KDE (as in KDEy) with a shrinkage-regularized log-ratio (CLR/Aitchison) kernel density estimator, derives the corresponding mixture MLE for point estimates, and obtains a Jacobian-free Bayesian posterior over prevalences. Shrinkage T_λ toward the barycenter plus bandwidth rescaling h_eff=(1-λ)h is justified by a local linearization of CLR (Proposition 1 and Appendices A–C). Experiments on 42 tabular/text/image datasets under Dirichlet bag sampling compare AE/W ranks and coverage/amplitude against CC/PCC/BBSE/EMQ/KDEy and Bayesian/bootstrap baselines, with ablations and Wilcoxon–Holm tests.

Significance. If the results hold, the work supplies a principled, practical geometry-aware continuous density model for multiclass quantification that both improves on Euclidean KDEy in many regimes (especially near the simplex boundary) and fills the missing Bayesian formulation for KDE-based quantifiers. Strengths include the clean change-of-variable argument that cancels the Jacobian for inference over π, the explicit local-linearization and bandwidth-scaling analysis, a large multi-domain evaluation with public code, and an ablation showing that shrinkage is necessary to avoid sharp failures of pure Aitchison KDE. The contribution is incremental relative to KDEy and CoDA kernels but is well-motivated and immediately usable for label-shift adaptation and uncertainty-aware prevalence estimation.

major comments (3)
  1. The central geometric claim is that the method respects simplex geometry rather than collapsing to Euclidean KDE. Proposition 1 and Appendix C show that as λ o1 the CLR map linearizes and the kernel becomes locally Euclidean (up to a constant). Model selection frequently chooses λ near 0 or near 1 (31% and 16.7% of datasets). The paper should quantify, for the selected λ values, how much the induced densities and prevalence estimates differ from a pure Euclidean KDEy with matched effective bandwidth; without that, the practical gain attributable to Aitchison geometry versus adaptive shrinkage remains only partially characterized (Tables 1–2 and D.5).
  2. Bayesian evaluation (Table 2, Tables 8–9) reports hard/soft coverage gaps and Monte-Carlo amplitude of the classwise hyper-rectangle. Amplitude is not a proper multivariate credible-region volume, and Bonferroni hard coverage is conservative; the ranking of methods can therefore be sensitive to the chosen uncertainty summary. A short sensitivity check with a proper joint credible region (or at least the average interval score already used for temperature selection) would strengthen the claim that Bayes-KDEy(Ait-λ) is among the strongest Bayesian quantifiers.
  3. The load-bearing modeling assumption is the classical label-shift invariance Ps(p|Y)=Pt(p|Y) (§3). The authors correctly flag this limitation and note covariate/sparse-joint shift as future work. Because every posterior-based baseline shares the same premise, it does not invalidate the geometry comparison, but the abstract and conclusions should state more explicitly that all reported gains are conditional on that invariance holding.
minor comments (4)
  1. Figure 1 is described but the actual visual comparison of Euclidean vs. geometry-aware kernels is only sketched; a clearer panel or caption would help readers who are not already familiar with Aitchison geometry.
  2. Notation for the tempered posterior (T) and the shrinkage parameter (λ) is introduced cleanly, but the relationship between the validation objective (AE) used for h,λ and the Winkler score used for T could be stated once in the main text rather than only in Appendix D.2.
  3. A few typographical inconsistencies appear (e.g., “Bayesian Quantification via Compositional Data Analysis” header vs. title casing; occasional missing spaces around citations). These are easily fixed.
  4. The claim that Dirichlet kernels lack drop-in ML implementations is fair, but a brief pointer to any recent software (or an explicit statement that none was usable for the multiclass posterior setting) would complete the related-work discussion.

Circularity Check

0 steps flagged · score 1.0 of 10

No load-bearing circularity: mixture-MLE + CoDA geometry derivation is self-contained; self-citations are scaffolding/baselines only.

full rationale

The central derivation begins from the standard KL/MLE mixture view of quantification (Remark 1 / Eq. 1), places class-conditional densities in CLR coordinates after a fixed shrinkage map T_λ (Eq. 2), and notes that the Jacobian of ϕ∘T_λ is independent of π so both the point estimator and the unnormalized Bayesian posterior (Eq. 3) may be written in transformed coordinates. Local linearization (Prop. 1) and bandwidth rescaling heff=(1-λ)h are proved in the appendices from first-order Taylor expansion and affine contraction; they do not presuppose the empirical ranking. Hyper-parameters h, λ and temperature T are chosen by ordinary validation on held-out bags and then evaluated on independent Dirichlet-sampled test bags drawn from 42 external datasets; the reported AE/W ranks and coverage gaps are therefore not forced by construction. Self-citations to the authors’ prior KDEy and QuaPy supply the Euclidean baseline and software framework, not a uniqueness theorem or an ansatz that closes the logical loop. The only structural premise is classical label-shift invariance of class-conditional posteriors, which is external and already flagged as a limitation. Consequently the paper exhibits at most the mild, non-load-bearing self-citation typical of incremental methodological work.

Assumptions & free parameters 3 free parameters · 5 assumptions · 2 invented entities

The central claim rests on standard label-shift and posterior-invariance assumptions from the quantification literature, plus ordinary KDE/MLE machinery. What the paper adds is the shrinkage map Tλ, bandwidth rescaling, and the Bayesian tempered posterior over mixture weights—not new physical entities. Free parameters are the usual bandwidth, shrinkage, and temperature selected on validation bags.

free parameters (3)
  • shrinkage λ = dataset-specific; most frequent 0.001 (31%), second 0.999 (16.7%)
    Chosen per dataset from {0.001,0.25,0.5,0.75,0.9,0.999} by minimizing AE on 100 Dirichlet(1) validation bags; controls how close the estimator is to pure Aitchison vs near-Euclidean.
  • kernel bandwidth h = dataset-specific via validation
    Selected on the same validation bags (logspace grids differ for Gaussian vs Aitchison); effective bandwidth heff=(1−λ)h is derived, but base h is free.
  • posterior temperature T = dataset-specific after model selection
    Post-hoc calibrated on validation bags by mean Winkler interval score over grid including 0.5,1,1.5,2,5,10,100,1000; controls posterior concentration under misspecification.
assumptions (5)
  • domain assumption Label (prior probability) shift: Ps(Y)≠Pt(Y) but Ps(X|Y)=Pt(X|Y), implying class-conditional posterior laws are domain-invariant.
    Stated in §3 Problem Setup; entire quantification-in-posterior-space pipeline depends on it.
  • domain assumption Quantification as mixture density matching via KL/MLE: π̂ = argmax_π Σ log m_π(p^(i)).
    Remark 1 and Eq. (1); standard for MLLS/KDEy-style methods.
  • standard math CLR/ILR maps equip the simplex with Aitchison geometry so Gaussian KDE in transformed space is geometry-aware.
    §5; classical CoDA (Aitchison); ILR isometry of CLR used for convenience.
  • ad hoc to paper Affine shrinkage Tλ(p)=(1−λ)p+λu with heff=(1−λ)h preserves relative smoothing and linearizes CLR near the barycenter.
    Proposition 1 and Appendices A–C; the specific regularizer and bandwidth rule are paper-specific design choices.
  • standard math Jacobian of ϕ∘Tλ is π-independent, so likelihood/posterior for π can ignore it.
    §5.1–5.2; change-of-variables fact used to make Bayesian KDE practical.
invented entities (2)
  • Shrinkage-regularized geometry-aware (Aitchison) KDE quantifier KDEy(Ait-λ)
    purpose: Estimate class-conditional densities of posteriors on the simplex without Euclidean mass leakage, with stability near boundaries.
    Methodological construct, not a latent physical object; independent evidence is empirical performance and ablation vs λ=0, not an external measurement of a new quantity.
  • Tempered Bayesian posterior for KDE mixture prevalences
    purpose: Provide uncertainty over π and control dispersion under model misspecification via temperature T.
    Standard tempering applied to the new KDE likelihood; falsifiable only via coverage/amplitude experiments in the paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Geometry-Aware Bayesian Quantification via Compositional Data Analysis." pith.science (2026). https://pith.science/paper/J632IS7T

@misc{pith2026260704977,
  author       = {Pith},
  title        = {Pith review of: Geometry-Aware Bayesian Quantification via Compositional Data Analysis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/J632IS7T}},
  note         = {Machine review of arXiv:2607.04977}
}
read the original abstract

Accurately estimating the unknown target label distribution is the critical first step for adapting to label shift. This task, widely known as quantification or class prevalence estimation, has recently seen significant advances through continuous KDE-based methods which model the density of multiclass classifier posteriors. Posterior vectors might be regarded as compositional data, since they lie on the probability simplex. However, existing KDE-based quantifiers typically rely on Euclidean Gaussian kernels, which ignore simplex geometry and incorrectly assign probability mass outside its boundaries. We introduce a geometry-aware KDE model for multiclass quantification based on log-ratio representations and Aitchison geometry, together with a shrinkage regularization that improves robustness near the simplex boundary. Combined with a maximum-likelihood interpretation of KDE-based quantification, we derive both point-estimation and Bayesian inference procedures for class prevalences. Experiments on 42 datasets across tabular, text, and image domains show that the proposed method is competitive with state-of-the-art quantifiers, often improving over standard KDE-based baselines, while also yielding strong results among Bayesian quantification methods.

Figures

Figures reproduced from arXiv: 2607.04977 by the authors.

Figure 1
Figure 1. Comparison between a standard Gaussian kernel in Euclidean space and a geometry-aware kernel induced [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Effect of shrinkage on a geometry-aware kernel centered in the simplex for [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 4
Figure 4. These settings favor prevalences near the simplex boundary and prevalences concentrated near the barycenter, [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figures from the paper (2 more)
Figure 3
Figure 3. Figure 3: Average AE ranks in low-, medium-, and high-shift regimes for the three data modalities. [PITH_FULL_IMAGE:figures/full_fig_p008_3.png]
Figure 4
Figure 4. Figure 4: Violin plots for AE score distributions (log scale) for different target distributions. [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

47 extracted references · 3 linked inside Pith

  1. [1]

    The statistical analysis of compositional data.Journal of the Royal Statistical Society: Series B (Methodological), 44(2):139–160, 1982

    John Aitchison. The statistical analysis of compositional data.Journal of the Royal Statistical Society: Series B (Methodological), 44(2):139–160, 1982

    1982

  2. [2]

    Kernel density estimation for compositional data.Journal of the Royal Statistical Society: Series C (Applied Statistics), 34(2):129–137, 1985

    John Aitchison and Ian J Lauder. Kernel density estimation for compositional data.Journal of the Royal Statistical Society: Series C (Applied Statistics), 34(2):129–137, 1985

    1985

  3. [3]

    Maximum likelihood with bias-corrected calibration is hard-to-beat at label shift adaptation

    Amr Alexandari, Anshul Kundaje, and Avanti Shrikumar. Maximum likelihood with bias-corrected calibration is hard-to-beat at label shift adaptation. InProceedings of the 37th International Conference on Machine Learning (ICML 2020), pages 222–232, Virtual Event, 2020

    2020

  4. [4]

    Regularized learning for domain adaptation under label shifts

    Kamyar Azizzadenesheli, Anqi Liu, Fanny Yang, and Animashree Anandkumar. Regularized learning for domain adaptation under label shifts. InProceedings of the 7th International Conference on Learning Representations (ICLR 2019), New Orleans, US, 2019

    2019

  5. [5]

    González, and Juan J

    Alberto Castaño, Pablo González, Jaime A. González, and Juan J. del Coz. Matching distributions algorithms based on the Earth Mover’s Distance for ordinal quantification.IEEE Transactions On Neural Networks and Learning Systems, 2022. Forthcoming

    2022

  6. [6]

    Gaussian kernels for density estimation with compositional data.Computers & Geosciences, 37(5):702–711, 2011

    José E Chacón, G Mateu-Figueras, and Josep-Antoni Martín-Fernández. Gaussian kernels for density estimation with compositional data.Computers & Geosciences, 37(5):702–711, 2011

    2011

  7. [7]

    Estimating and explaining model performance when both covariates and labels shift.Advances in Neural Information Processing Systems, 35:11467–11479, 2022

    Lingjiao Chen, Matei Zaharia, and James Y Zou. Estimating and explaining model performance when both covariates and labels shift.Advances in Neural Information Processing Systems, 35:11467–11479, 2022

    2022

  8. [8]

    Distribution matching for graph quantification under structural covariate shift

    Clemens Damke and Eyke Hüllermeier. Distribution matching for graph quantification under structural covariate shift. InJoint European Conference on Machine Learning and Knowledge Discovery in Databases, pages 403–419. Springer, 2025

    2025

Show all 47 references
  1. [9]

    Daughton and Michael J

    Ashlynn R. Daughton and Michael J. Paul. Constructing accurate confidence intervals when aggregating social media data for public health monitoring. InProceedings of the 3rd AAAI International Workshop on Health Intelligence (W3PHIAI 2019), pages 9–17, Phoenix, US, 2019

    2019

  2. [10]

    Gain-some-lose-some: Reliable quantification under general dataset shift

    Benjamin Denham, Edmund MK Lai, Roopak Sinha, and M Asif Naeem. Gain-some-lose-some: Reliable quantification under general dataset shift. In2021 IEEE International Conference on Data Mining (ICDM), pages 1048–1053. IEEE, 2021

    2021

  3. [11]

    Label shift quantification with robustness guarantees via distribution feature matching

    Bastien Dussap, Gilles Blanchard, and Badr-Eddine Chérief-Abdellatif. Label shift quantification with robustness guarantees via distribution feature matching. InProceedings of the 34th European Conference on Machine Learning and Principles and Practice of Knowledge Discovery i...

    2023

  4. [12]

    LeQua@CLEF 2022: Learning to quantify

    Andrea Esuli, Alejandro Moreo, and Fabrizio Sebastiani. LeQua@CLEF 2022: Learning to quantify. InEuropean Conference on Information Retrieval, pages 374–381. Springer, 2022

    2022

  5. [13]

    An overview of LeQua 2024, the 2nd international data challenge on learning to quantify

    Andrea Esuli, Alejandro Moreo, Fabrizio Sebastiani, and Gianluca Sperduti. An overview of LeQua 2024, the 2nd international data challenge on learning to quantify. In4th International Workshop on Learning to Quantify (LQ 2024), page 51, 2024. 9 Bayesian Quantification via Comp...

    2024

  6. [14]

    Quantification under prior probability shift: The ratio estimator and its extensions.Journal of Machine Learning Research, 20:79:1–79:33, 2019

    Afonso Fernandes Vaz, Rafael Izbicki, and Rafael Bassi Stern. Quantification under prior probability shift: The ratio estimator and its extensions.Journal of Machine Learning Research, 20:79:1–79:33, 2019

    2019

  7. [15]

    Generalized Bayes quantification learning under dataset shift.Journal of the American Statistical Association, 117(540):2163–2181, 2022

    Jacob Fiksel, Abhirup Datta, Agbessi Amouzou, and Scott Zeger. Generalized Bayes quantification learning under dataset shift.Journal of the American Statistical Association, 117(540):2163–2181, 2022

    2022

  8. [16]

    Counting positives accurately despite inaccurate classification

    George Forman. Counting positives accurately despite inaccurate classification. InProceedings of the 16th European Conference on Machine Learning (ECML 2005), pages 564–575, Porto, PT, 2005

    2005

  9. [17]

    From classification to quantification in Tweet sentiment analysis.Social Network Analysis and Mining, 6(1):19, 2016

    Wei Gao and Fabrizio Sebastiani. From classification to quantification in Tweet sentiment analysis.Social Network Analysis and Mining, 6(1):19, 2016

    2016

  10. [18]

    A unified view of label shift estimation

    Saurabh Garg, Yifan Wu, Sivaraman Balakrishnan, and Zachary Lipton. A unified view of label shift estimation. In Proceedings of the 34th Conference on Neural Information Processing Systems (NeurIPS 2020), pages 3290–3300, Virtual Event, 2020

    2020

  11. [19]

    Strictly proper scoring rules, prediction, and estimation.Journal of the American statistical Association, 102(477):359–378, 2007

    Tilmann Gneiting and Adrian E Raftery. Strictly proper scoring rules, prediction, and estimation.Journal of the American statistical Association, 102(477):359–378, 2007

    2007

  12. [20]

    Navigating prevalence shifts in image analysis algorithm deployment.Medical image analysis, page 103504, 2025

    Patrick Godau, Piotr Kalinowski, Evangelia Christodoulou, Annika Reinke, Minu Tizabi, Luciana Ferrer, Paul Jäger, and Lena Maier-Hein. Navigating prevalence shifts in image analysis algorithm deployment.Medical image analysis, page 103504, 2025

    2025

  13. [21]

    Class distribution estimation based on the Hellinger distance.Information Sciences, 218:146–164, 2013

    Víctor González-Castro, Rocío Alaiz-Rodríguez, and Enrique Alegre. Class distribution estimation based on the Hellinger distance.Information Sciences, 218:146–164, 2013

    2013

  14. [22]

    On calibration of modern neural networks

    Chuan Guo, Geoff Pleiss, Yu Sun, and Kilian Q Weinberger. On calibration of modern neural networks. In International conference on machine learning, pages 1321–1330. PMLR, 2017

    2017

  15. [23]

    The No-U-Turn sampler: adaptively setting path lengths in hamiltonian monte carlo.J

    Matthew D Hoffman, Andrew Gelman, et al. The No-U-Turn sampler: adaptively setting path lengths in hamiltonian monte carlo.J. Mach. Learn. Res., 15(1):1593–1623, 2014

    2014

  16. [24]

    Hopkins and Gary King

    Daniel J. Hopkins and Gary King. A method of automated nonparametric content analysis for social science. American Journal of Political Science, 54(1):229–247, 2010

    2010

  17. [25]

    Bayesian-based online label shift estimation with dynamic Dirichlet priors.arXiv preprint arXiv:2511.18615, 2025

    Jiawei Hu and Javier A Barria. Bayesian-based online label shift estimation with dynamic Dirichlet priors.arXiv preprint arXiv:2511.18615, 2025

    2025

  18. [26]

    Estimating prevalence with precision and accuracy

    Aime Bienfait Igiraneza, Christophe Fraser, and Robert Hinch. Estimating prevalence with precision and accuracy. arXiv preprint arXiv:2507.06061, 2025

    2025 arXiv

  19. [27]

    Quantifying query fairness under unawareness.Journal of Artificial Intelligence Research, 85, 2026

    Thomas Jaenich, Alejandro Moreo, Alessandro Fabris, Graham McDonald, Andrea Esuli, Iadh Ounis, and Fabrizio Sebastiani. Quantifying query fairness under unawareness.Journal of Artificial Intelligence Research, 85, 2026

    2026

  20. [28]

    The UCI machine learning repository

    Markelle Kelly, Rachel Longjohn, and Kolby Nottingham. The UCI machine learning repository. https: //archive.ics.uci.edu

  21. [29]

    Learning multiple layers of features from tiny images

    Alex Krizhevsky and Geoffrey Hinton. Learning multiple layers of features from tiny images. Technical report, University of Toronto, Toronto, Ontario, 2009

    2009

  22. [30]

    Yann LeCun, Corinna Cortes, and Christopher J. C. Burges. The MNIST database of handwritten digits. http://yann.lecun.com/exdb/mnist/, 1998

    1998

  23. [31]

    Lipton, Yu-Xiang Wang, and Alexander J

    Zachary C. Lipton, Yu-Xiang Wang, and Alexander J. Smola. Detecting and correcting for label shift with black box predictors. InProceedings of the 35th International Conference on Machine Learning (ICML 2018), pages 3128–3136, Stockholm, SE, 2018

    2018

  24. [32]

    Updating on the kernel density estimation for compo- sitional data

    JA Martın-Fernández, JE Chacón, and G Mateu-Figueras. Updating on the kernel density estimation for compo- sitional data. Inof: Proceedings of the 17th Conference IASC-ERSS, Compstat, Roma,(Italy), pages 713–720, 2006

    2006

  25. [33]

    Efficient quantification on large-scale networks.Machine Learning, 114(12):1–26, 2025

    Alessio Micheli, Alejandro Moreo, Marco Podda, Fabrizio Sebastiani, William Simoni, and Domenico Tortorella. Efficient quantification on large-scale networks.Machine Learning, 114(12):1–26, 2025

    2025

  26. [34]

    QuaPy: a Python-based framework for quantification

    Alejandro Moreo, Andrea Esuli, and Fabrizio Sebastiani. QuaPy: a Python-based framework for quantification. InProceedings of the 30th ACM International Conference on Information & Knowledge Management, pages 4534–4543, 2021

    2021

  27. [35]

    Kernel density estimation for multiclass quantification

    Alejandro Moreo, Pablo González, and Juan José del Coz. Kernel density estimation for multiclass quantification. Machine Learning, 114(4), 2025

    2025

  28. [36]

    Reading digits in natural images with unsupervised feature learning

    Yuval Netzer, Tao Wang, Adam Coates, Alessandro Bissacco, Baolin Wu, Andrew Y Ng, et al. Reading digits in natural images with unsupervised feature learning. InNIPS workshop on deep learning and unsupervised feature learning, volume 2011, page 4. Granada, 2011. 10 Bayesian Qua...

    2011

  29. [37]

    Asymptotic properties of Dirichlet kernel density estimators

    Frédéric Ouimet and Raimon Tolosana-Delgado. Asymptotic properties of Dirichlet kernel density estimators. Journal of Multivariate Analysis, 187:104832, 2022

    2022

  30. [38]

    Adjusting the outputs of a classifier to new a priori probabilities: A simple procedure.Neural Computation, 14(1):21–41, 2002

    Marco Saerens, Patrice Latinne, and Christine Decaestecker. Adjusting the outputs of a classifier to new a priori probabilities: A simple procedure.Neural Computation, 14(1):21–41, 2002

    2002

  31. [39]

    A comparative evaluation of quantification methods.Journal of Machine Learning Research, 26(55):1–54, 2025

    Tobias Schumacher, Markus Strohmaier, and Florian Lemmerich. A comparative evaluation of quantification methods.Journal of Machine Learning Research, 26(55):1–54, 2025

    2025

  32. [40]

    Evaluation measures for quantification: An axiomatic approach.Information Retrieval Journal, 23(3):255–288, 2020

    Fabrizio Sebastiani. Evaluation measures for quantification: An axiomatic approach.Information Retrieval Journal, 23(3):255–288, 2020

    2020

  33. [41]

    When training and test sets are different: Characterizing learning transfer

    Amos Storkey. When training and test sets are different: Characterizing learning transfer. In Joaquin Quiñonero- Candela, Masashi Sugiyama, Anton Schwaighofer, and Neil D. Lawrence, editors,Dataset shift in machine learning, pages 3–28. The MIT Press, Cambridge, US, 2009

    2009

  34. [42]

    Confidence intervals for class prevalences under prior probability shift.Machine Learning and Knowledge Extraction, 1(3):805–831, 2019

    Dirk Tasche. Confidence intervals for class prevalences under prior probability shift.Machine Learning and Knowledge Extraction, 1(3):805–831, 2019

    2019

  35. [43]

    Class prior estimation under covariate shift: No problem?arXiv preprint arXiv:2206.02449, 2022

    Dirk Tasche. Class prior estimation under covariate shift: No problem?arXiv preprint arXiv:2206.02449, 2022

    2022 arXiv

  36. [44]

    LEAP: Linear equations for classifier accuracy prediction under prior probability shift.Machine Learning, 114(12):293, 2025

    Lorenzo V olpi, Alejandro Moreo, and Fabrizio Sebastiani. LEAP: Linear equations for classifier accuracy prediction under prior probability shift.Machine Learning, 114(12):293, 2025

    2025

  37. [45]

    Fashion-MNIST: a novel image dataset for benchmarking machine learning algorithms.arXiv preprint arXiv:1708.07747, 2017

    Han Xiao, Kashif Rasul, and Roland V ollgraf. Fashion-MNIST: a novel image dataset for benchmarking machine learning algorithms.arXiv preprint arXiv:1708.07747, 2017

    2017 arXiv

  38. [46]

    Label shift estimation for class-imbalance problem: A Bayesian approach

    Changkun Ye, Russell Tsuchida, Lars Petersson, and Nick Barnes. Label shift estimation for class-imbalance problem: A Bayesian approach. InProceedings of the IEEE/CVF winter conference on applications of computer vision, pages 1073–1082, 2024

    2024

  39. [47]

    Bayesian quantification with black-box estimators.Transactions on Machine Learning Research, 2024

    Albert Ziegler and Paweł Czy˙z. Bayesian quantification with black-box estimators.Transactions on Machine Learning Research, 2024. 11 Bayesian Quantification via Compositional Data Analysis A. Moreo, P. González, JJ. del Coz Appendix A Proof of Proposition 1: linearization of ...

    2024

Pith tools

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