pith. sign in

arxiv: 2604.17236 · v1 · submitted 2026-04-19 · 🧮 math.ST · stat.TH

Learning Mixtures of Nonparametric and Convolutional Measures on Effectively Low-dimensional Affine Spaces

Sunrit Chakraborty , XuanLong Nguyen This is my paper

Pith reviewed 2026-05-10 06:12 UTC · model grok-4.3

classification 🧮 math.ST stat.TH
keywords mixture modelsidentifiabilitysubspace clusteringnonparametric statisticsBayesian inferenceconvolutional measureslow-dimensional subspacesspectral unmixing
0
0 comments X

The pith

Finite mixtures of convolutional measures on low-dimensional affine subspaces have uniquely identifiable minimal representations in semi-parametric settings.

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

The paper develops a model for data distributed near mixtures of low-dimensional affine subspaces using finite mixtures in which each component is a distribution supported on such a subspace convolved with a noise kernel. It proves that these mixtures are identifiable by showing that the minimal representation is uniquely recoverable from the observed distribution under very general conditions, exploiting the geometric structure of the supports. The work also establishes posterior contraction rates for the parameters in a Bayesian setting when the supports are restricted to convex polytopes, which requires new inverse bounds for the resulting nested mixture problem. A reader would care because the result supplies theoretical conditions under which multiple latent low-dimensional structures can be learned from noisy continuous data. This directly grounds methods for subspace clustering and related tasks such as spectral unmixing.

Core claim

The central claim is that the minimal representation for finite mixtures of nonparametric and convolutional measures on low-dimensional affine spaces is uniquely identifiable in a semi-parametric setting. Each component arises from convolving a distribution supported on a low-dimensional subspace with a suitable noise kernel, and identifiability follows from the geometric structure of these supports. For a parametrized subclass in which the component supports are convex polytopes, posterior contraction rates are derived in a well-specified Bayesian regime, relying on novel inverse bounds that handle the nested continuous mixture structure inside the outer mixture kernel.

What carries the argument

The geometric structure of the supports of the latent measures on low-dimensional affine subspaces, which separates the convolutional components and yields unique minimal representations of the overall mixture.

If this is right

  • The component mixing measures and their low-dimensional supports can be uniquely recovered from the observed mixture distribution.
  • Posterior distributions contract around the true parameters at explicit rates when supports are convex polytopes under a well-specified Bayesian model.
  • New inverse bounds are obtained for nested mixtures in which the mixing kernel itself is a continuous mixture.
  • The framework supplies conditions for learning multiple latent low-dimensional structures via subspace clustering.
  • The identifiability theory extends to applications such as end-member analysis, spectral unmixing, and topic models.

Where Pith is reading between the lines

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

  • The geometric approach may generalize to cases where the noise kernel is learned from data rather than treated as known.
  • Connections to manifold learning suggest that the same support geometry could be used to test whether observed data truly concentrate near affine subspaces versus curved manifolds.
  • A direct empirical test would be to apply the developed algorithms to benchmark subspace-clustering datasets and measure recovery error as dimension or noise level varies.

Load-bearing premise

The observations are i.i.d. draws from a mixture in which each component is the convolution of a distribution supported on a low-dimensional affine subspace with a noise kernel.

What would settle it

Construct two distinct minimal mixtures of such convolutional measures that generate exactly the same observed distribution; if such pairs exist, the unique-identifiability claim is false.

Figures

Figures reproduced from arXiv: 2604.17236 by Sunrit Chakraborty, XuanLong Nguyen.

Figure 1
Figure 1. Figure 1: Example of model in 𝐷 = 3 (for visualization) with 𝐾 = 3 components: two 2−dimensional components 𝐺1, 𝐺2 with supports S1,S2 shown in orange and teal colors (intersecting along the 1−dimensional line segment shown with dashed line), and one 1−dimensional component 𝐺3, whose support is the union of two line segments on the common affine space (a line), shown in red. The mea￾sure 𝐺1 (orange) is supported on … view at source ↗
Figure 2
Figure 2. Figure 2: Example of model in 𝐷 = 3 (for visualization) with 𝐾 = 2 components, each latent measure 𝐺𝑘 is supported on a 2-dimensional polytope (triangle here) with the dashed line showing the intersec￾tion of the supports. Left panel shows the noiseless case, while the right panel shows a scatter plot of observations from from the model with Gaussian noise. The underlying 𝜇𝑘 is Dirichlet and its effect on 𝐺𝑘 can be … view at source ↗
Figure 3
Figure 3. Figure 3: Examples in R 2 illustrating total exposure definition: Examples (a) and (b) satisfy A, but neither are totally exposed In (a), two of the three polytopes are exposed, while no polytope is exposed in (b). Example (c) is totally exposed but does not satisfy A. Note when ambient dimension 𝐷 is large and component polytopes are in general position, they almost surely satisfy both A and totally exposed. parame… view at source ↗
Figure 4
Figure 4. Figure 4: Simulation Results in a Single Component Setting [PITH_FULL_IMAGE:figures/full_fig_p024_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Simulation Results in General Setting 4000 and consider 50 repeated experiments for setting, using 5 types of algorithms as discussed in the previous section, tabulated below. The performance of the algorithms is measured in terms of the metric 𝑑 defined in Equation (11) In Setting 1, we set 𝐾 = 3, 𝑑 = 2, 𝐷 = 3 where each component is a line-segment in three-dimensions. However, the ground-truth components… view at source ↗
Figure 6
Figure 6. Figure 6: Results of Model Selection using BIC for Setting 2 [PITH_FULL_IMAGE:figures/full_fig_p026_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Illustration of the proof: Starting from [PITH_FULL_IMAGE:figures/full_fig_p040_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Extracting the effect of coefficient 𝑎11 corresponding to vertex 𝜃11 in the proof of the inverse bound (case 𝐾 = 2, 𝑑 = 3, 𝐷 = 3: first a change of coordinates from (𝑋1, 𝑋2, 𝑋3) to (𝑋˜ 1, 𝑋˜ 2, 𝑋˜ 3) using translation and orthogonal rotation only – in this new system 𝜃𝑘 𝑗 becomes 𝜃˜ 𝑘 𝑗 and 𝑎𝑘 𝑗 becomes 𝑎˜𝑘 𝑗. If 𝑎11 ≠ 0, such a coordinate change is possible ensuring 𝑎˜111 ≠ 0. Vertex 𝜃11 is an exposed poi… view at source ↗
Figure 9
Figure 9. Figure 9: Settings for single component simulations in Section [PITH_FULL_IMAGE:figures/full_fig_p053_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Settings for Simulations in Section 5.2 for multiple components - visualization via PCA using first 2 principle components Algorithm 𝑛 = 200 𝑛 ≈ 1000 𝑛 = 4000 Gaussian 0.81 1.47 1.64 MCMC 7.60 7.83 9.46 EM(50) 0.55 0.62 0.69 EM(100) 0.56 0.65 0.66 EM (400) 0.67 0.69 0.76 [PITH_FULL_IMAGE:figures/full_fig_p054_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: (Left) Simulation results for Setting 3, (Right) Illustration of a Local Model in Setting 1 [PITH_FULL_IMAGE:figures/full_fig_p055_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Result of Using Approximate EM Algorithm (assuming Dirichlet latent mixing) when the [PITH_FULL_IMAGE:figures/full_fig_p055_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Results of using Approximate EM algorithm when the latent measure is mis-specified [PITH_FULL_IMAGE:figures/full_fig_p056_13.png] view at source ↗
read the original abstract

In this paper, we develop a finite mixture of convolutional distributions, a statistical model to analyze continuous data distributed approximately on a mixture of low-dimensional affine subspaces. The observations are assumed independent and identically distributed from the mixture of distributions, where each component arises from a convolution of a distribution supported on a low-dimensional subspace with a suitable noise kernel. We discuss theoretical properties of such class of models, including identifiability under very general conditions - in particular, showing that the minimal representation for such mixtures is uniquely identifiable in a semi-parametric setting. We further study the posterior contraction rates for the parameters for a parametrized class of such models where the supports of the component mixing measures are assumed to be convex polytopes under a suitable well-specified Bayesian regime. This still requires developing novel inverse bounds for problems involving a nested mixture structure, where the mixture kernel is itself another continuous mixture. Our approach for both the identifiability theory and posterior contraction rates is to exploit the geometric structure of the underlying support of the latent measures. Apart from applications in end-member analysis, spectral unmixing and topic models, this study provides a grounded framework for subspace clustering with the goal of exploring conditions for learning multiple latent low-dimensional structures. We illustrate our findings through careful simulation study, which also includes developing new algorithms for such class of models

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 manuscript develops a finite mixture model of convolutional distributions for continuous data approximately supported on mixtures of low-dimensional affine subspaces. Each component arises from convolving a nonparametric measure on a low-dimensional affine support with a noise kernel. The central claims are that the minimal representation of such mixtures is uniquely identifiable in a semi-parametric setting, and that posterior contraction rates can be derived for a parametrized subclass where the latent supports are convex polytopes, under a well-specified Bayesian regime. Both results exploit geometric properties of the supports; the paper also presents simulation studies and associated algorithms.

Significance. If the identifiability and contraction-rate results are rigorously established, the work supplies a useful theoretical framework for subspace clustering and related inverse problems (spectral unmixing, end-member analysis, topic models). The geometric approach to handling nested mixtures and the derivation of inverse bounds for the convolution structure represent a clear advance over standard mixture theory. The simulation component, while secondary, helps ground the claims.

major comments (2)
  1. [Identifiability section] § on identifiability (semi-parametric setting): the uniqueness argument for the minimal representation relies on general geometric conditions on the affine supports and the noise kernel; however, it is not shown whether these conditions remain sufficient when the number of mixture components is unknown or when the noise kernel itself belongs to a nonparametric class, which is load-bearing for the semi-parametric claim.
  2. [Posterior contraction rates section] § on posterior contraction rates (convex-polytope case): the novel inverse bounds for the nested mixture (outer mixture of convolutions, inner mixture over the polytope support) are central to obtaining the stated rates; the manuscript does not provide an explicit comparison of these rates to the minimax rates for ordinary finite mixtures or to the rates that would hold without the low-dimensional affine assumption, making it difficult to assess the improvement attributable to the geometric structure.
minor comments (2)
  1. [Abstract / Introduction] The abstract and introduction refer to 'a parametrized class of such models' without immediately defining the parametrization; a short clarifying sentence or reference to the relevant section would improve readability.
  2. [Simulation study] Simulation study: the description of the new algorithms is brief; adding pseudocode or a high-level complexity statement would help readers reproduce the numerical results.

Simulated Author's Rebuttal

2 responses · 0 unresolved

Thank you for the opportunity to respond to the referee's report. We appreciate the referee's recognition of the potential utility of our framework for subspace clustering and related inverse problems. Below we provide point-by-point responses to the major comments, indicating where revisions will be made to the manuscript.

read point-by-point responses

  1. Referee: [Identifiability section] § on identifiability (semi-parametric setting): the uniqueness argument for the minimal representation relies on general geometric conditions on the affine supports and the noise kernel; however, it is not shown whether these conditions remain sufficient when the number of mixture components is unknown or when the noise kernel itself belongs to a nonparametric class, which is load-bearing for the semi-parametric claim.

    Authors: The uniqueness result is stated for the minimal representation, which by definition corresponds to the smallest number of components necessary to represent the mixture; thus, it inherently applies when the number of components is unknown. The geometric conditions on the supports are used to establish this uniqueness. In our semi-parametric model, the noise kernel is taken to be fixed and known, while the nonparametric components are the mixing measures supported on the affine spaces. We will revise the manuscript to explicitly state these assumptions and add a discussion on the scope of the semi-parametric claim, including why extending to a nonparametric kernel would fall outside the current framework. revision: partial

  2. Referee: [Posterior contraction rates section] § on posterior contraction rates (convex-polytope case): the novel inverse bounds for the nested mixture (outer mixture of convolutions, inner mixture over the polytope support) are central to obtaining the stated rates; the manuscript does not provide an explicit comparison of these rates to the minimax rates for ordinary finite mixtures or to the rates that would hold without the low-dimensional affine assumption, making it difficult to assess the improvement attributable to the geometric structure.

    Authors: We acknowledge that an explicit comparison would strengthen the presentation. In the revised version, we will add a subsection discussing the obtained contraction rates in relation to standard minimax rates for finite mixtures in high dimensions (e.g., those depending on the ambient dimension) and contrast them with the rates that exploit the low-dimensional affine structure, thereby clarifying the improvement due to the geometric assumptions. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper's central results on semi-parametric identifiability of minimal representations and posterior contraction rates for mixtures of convolutional measures on low-dimensional affine subspaces rely on explicit i.i.d. sampling assumptions, geometric properties of convex polytope supports, and standard Bayesian well-specified regimes. These are stated as modeling primitives rather than derived from fitted quantities or self-referential definitions. No load-bearing step reduces a prediction to an input by construction, invokes self-citation for uniqueness theorems, or renames known results; the derivation chain remains self-contained against external geometric and statistical benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 3 axioms · 0 invented entities

The central claims rest on standard mixture modeling assumptions and geometric properties of subspaces, with no free parameters or invented entities explicitly introduced in the abstract.

axioms (3)
  • domain assumption Observations are i.i.d. from the mixture of convolutional distributions
    Explicitly stated as the data-generating assumption in the abstract.
  • domain assumption Supports of component mixing measures are convex polytopes
    Assumed for the parametrized class when studying posterior contraction rates.
  • domain assumption Suitable noise kernel for the convolution
    Required for each component to arise from convolution with a low-dimensional support.

pith-pipeline@v0.9.0 · 5531 in / 1369 out tokens · 47800 ms · 2026-05-10T06:12:56.016805+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

110 extracted references · 110 canonical work pages

  1. [1]

    2000 , publisher=

    Asymptotic statistics , author=. 2000 , publisher=

    work page 2000

  2. [2]

    arXiv preprint arXiv:1905.11009 , year=

    Dirichlet simplex nest and geometric inference , author=. arXiv preprint arXiv:1905.11009 , year=

    work page arXiv 1905

  3. [3]

    , booktitle=

    Gruber, Peter and Theis, Fabian J. , booktitle=. Grassmann clustering , year=

    work page

  4. [4]

    Advances in neural information processing systems , volume=

    A spectral algorithm for latent dirichlet allocation , author=. Advances in neural information processing systems , volume=

    work page

  5. [5]

    arXiv preprint arXiv:1710.11070 , year=

    Convergence Rates of Latent Topic Models Under Relaxed Identifiability Conditions , author=. arXiv preprint arXiv:1710.11070 , year=

    work page arXiv

  6. [6]

    2012 IEEE 53rd annual symposium on foundations of computer science , pages=

    Learning topic models--going beyond SVD , author=. 2012 IEEE 53rd annual symposium on foundations of computer science , pages=. 2012 , organization=

    work page 2012

  7. [7]

    International conference on machine learning , pages=

    A practical algorithm for topic modeling with provable guarantees , author=. International conference on machine learning , pages=. 2013 , organization=

    work page 2013

  8. [8]

    Bernoulli , volume=

    Posterior contraction of the population polytope in finite admixture models , author=. Bernoulli , volume=. 2015 , publisher=

    work page 2015

  9. [9]

    The Annals of Statistics , volume=

    Convergence of latent mixing measures in finite and infinite mixture models , author=. The Annals of Statistics , volume=. 2013 , publisher=

    work page 2013

  10. [10]

    Advances in Neural Information Processing Systems , volume=

    Geometric Dirichlet means algorithm for topic inference , author=. Advances in Neural Information Processing Systems , volume=

    work page

  11. [11]

    Gaussian LDA for topic models with word embeddings , author=. Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) , pages=

    work page

  12. [12]

    Chinese Conference on Pattern Recognition , pages=

    Latent topic model based on Gaussian-LDA for audio retrieval , author=. Chinese Conference on Pattern Recognition , pages=. 2012 , organization=

    work page 2012

  13. [13]

    Advances in neural information processing systems , volume=

    Hierarchical topic models and the nested Chinese restaurant process , author=. Advances in neural information processing systems , volume=

    work page

  14. [14]

    Journal of machine Learning research , volume=

    Latent dirichlet allocation , author=. Journal of machine Learning research , volume=

    work page

  15. [15]

    SIAM Journal on Matrix Analysis and Applications , volume=

    Schubert varieties and distances between subspaces of different dimensions , author=. SIAM Journal on Matrix Analysis and Applications , volume=. 2016 , publisher=

    work page 2016

  16. [16]

    Foundations of Computational Mathematics , volume=

    The Grassmannian of affine subspaces , author=. Foundations of Computational Mathematics , volume=. 2021 , publisher=

    work page 2021

  17. [17]

    The Annals of Mathematical Statistics , volume=

    Asymptotic properties of non-linear least squares estimators , author=. The Annals of Mathematical Statistics , volume=. 1969 , publisher=

    work page 1969

  18. [18]

    Annals of Statistics , pages=

    Convergence rates of posterior distributions , author=. Annals of Statistics , pages=. 2000 , publisher=

    work page 2000

  19. [19]

    The Annals of Statistics , pages=

    Probability inequalities for likelihood ratios and convergence rates of sieve MLEs , author=. The Annals of Statistics , pages=. 1995 , publisher=

    work page 1995

  20. [20]

    2017 , publisher=

    Fundamentals of nonparametric Bayesian inference , author=. 2017 , publisher=

    work page 2017

  21. [21]

    Journal of the Royal Statistical Society: Series B (Methodological) , volume=

    The statistical analysis of compositional data , author=. Journal of the Royal Statistical Society: Series B (Methodological) , volume=. 1982 , publisher=

    work page 1982

  22. [22]

    Journal of the Royal Statistical Society: Series C (Applied Statistics) , volume=

    The resolution of a compositional data set into mixtures of fixed source compositions , author=. Journal of the Royal Statistical Society: Series C (Applied Statistics) , volume=. 1993 , publisher=

    work page 1993

  23. [23]

    Proceedings of IGARSS'94-1994 IEEE International Geoscience and Remote Sensing Symposium , volume=

    Geometric mixture analysis of imaging spectrometry data , author=. Proceedings of IGARSS'94-1994 IEEE International Geoscience and Remote Sensing Symposium , volume=. 1994 , organization=

    work page 1994

  24. [24]

    JPL, Summaries of the 4th Annual JPL Airborne Geoscience Workshop

    Objective determination of image end-members in spectral mixture analysis of AVIRIS data , author=. JPL, Summaries of the 4th Annual JPL Airborne Geoscience Workshop. Volume 1: AVIRIS Workshop , year=

    work page

  25. [25]

    Mathematical Geosciences , volume=

    BEMMA: a hierarchical Bayesian end-member modeling analysis of sediment grain-size distributions , author=. Mathematical Geosciences , volume=. 2016 , publisher=

    work page 2016

  26. [26]

    Genetics , volume=

    Inference of population structure using multilocus genotype data , author=. Genetics , volume=. 2000 , publisher=

    work page 2000

  27. [27]

    Proceedings of the 22nd annual international ACM SIGIR conference on Research and development in information retrieval , pages=

    Probabilistic latent semantic indexing , author=. Proceedings of the 22nd annual international ACM SIGIR conference on Research and development in information retrieval , pages=

    work page

  28. [28]

    Proceedings of the SIGCHI conference on Human factors in computing systems , pages=

    Using latent semantic analysis to improve access to textual information , author=. Proceedings of the SIGCHI conference on Human factors in computing systems , pages=

    work page

  29. [29]

    Neural networks , volume=

    Independent component analysis: algorithms and applications , author=. Neural networks , volume=. 2000 , publisher=

    work page 2000

  30. [30]

    Nature , volume=

    Learning the parts of objects by non-negative matrix factorization , author=. Nature , volume=. 1999 , publisher=

    work page 1999

  31. [31]

    Advances in neural information processing systems , volume=

    Correlated topic models , author=. Advances in neural information processing systems , volume=. 2006 , publisher=

    work page 2006

  32. [32]

    A Hierarchical Bayesian Model for the Unmixing Analysis of Compositional Data subject to Unit-sum Constraints , author=

    work page

  33. [33]

    International Conference on Machine Learning , pages=

    Near-optimal sample complexity bounds for learning Latent k- polytopes and applications to Ad-Mixtures , author=. International Conference on Machine Learning , pages=. 2020 , organization=

    work page 2020

  34. [34]

    arXiv preprint arXiv:2002.10855 , year=

    Gaussian hierarchical latent dirichlet allocation: bringing polysemy back , author=. arXiv preprint arXiv:2002.10855 , year=

    work page arXiv 2002

  35. [35]

    Journal of Classification , pages=

    Chimeral Clustering , author=. Journal of Classification , pages=. 2021 , publisher=

    work page 2021

  36. [36]

    The Annals of Statistics , pages=

    Optimal rate of convergence for finite mixture models , author=. The Annals of Statistics , pages=. 1995 , publisher=

    work page 1995

  37. [37]

    Journal of the Royal Statistical Society: Series B (Statistical Methodology) , volume=

    Asymptotic behaviour of the posterior distribution in overfitted mixture models , author=. Journal of the Royal Statistical Society: Series B (Statistical Methodology) , volume=. 2011 , publisher=

    work page 2011

  38. [38]

    Journal of the American Statistical Association , volume=

    Bayesian model selection in finite mixtures by marginal density decompositions , author=. Journal of the American Statistical Association , volume=. 2001 , publisher=

    work page 2001

  39. [39]

    Electronic Journal of Statistics , volume=

    On strong identifiability and convergence rates of parameter estimation in finite mixtures , author=. Electronic Journal of Statistics , volume=. 2016 , publisher=

    work page 2016

  40. [40]

    Bernoulli , volume=

    On posterior contraction of parameters and interpretability in Bayesian mixture modeling , author=. Bernoulli , volume=. 2021 , publisher=

    work page 2021

  41. [41]

    arXiv preprint arXiv:2004.05542 , year=

    Convergence of de Finetti's mixing measure in latent structure models for observed exchangeable sequences , author=. arXiv preprint arXiv:2004.05542 , year=

    work page arXiv 2004

  42. [42]

    International Conference on Machine Learning , pages=

    Understanding the limiting factors of topic modeling via posterior contraction analysis , author=. International Conference on Machine Learning , pages=. 2014 , organization=

    work page 2014

  43. [43]

    International Conference on Machine Learning , pages=

    Provable algorithms for inference in topic models , author=. International Conference on Machine Learning , pages=. 2016 , organization=

    work page 2016

  44. [44]

    Electronic Journal of Statistics , volume=

    Convergence rates of latent topic models under relaxed identifiability conditions , author=. Electronic Journal of Statistics , volume=. 2019 , publisher=

    work page 2019

  45. [45]

    Journal of Multivariate Analysis , volume=

    A characterization of Dirichlet distributions , author=. Journal of Multivariate Analysis , volume=. 1988 , publisher=

    work page 1988

  46. [46]

    2005 , publisher=

    Testing statistical hypotheses , author=. 2005 , publisher=

    work page 2005

  47. [47]

    Bernoulli , number =

    Borrowing strengh in hierarchical Bayes: Posterior concentration of the Dirichlet base measure , urldate =. Bernoulli , number =

    work page

  48. [48]

    The Annals of Mathematical Statistics , volume=

    Identifiability of mixtures of product measures , author=. The Annals of Mathematical Statistics , volume=. 1967 , publisher=

    work page 1967

  49. [49]

    arXiv preprint arXiv:1807.05444 , year=

    On the identifiability of finite mixtures of finite product measures , author=. arXiv preprint arXiv:1807.05444 , year=

    work page arXiv

  50. [50]

    The Annals of Statistics , volume=

    An operator theoretic approach to nonparametric mixture models , author=. The Annals of Statistics , volume=. 2019 , publisher=

    work page 2019

  51. [51]

    The Annals of Probability , pages=

    Identifiability of continuous mixtures of unknown Gaussian distributions , author=. The Annals of Probability , pages=. 1985 , publisher=

    work page 1985

  52. [52]

    The annals of Mathematical statistics , volume=

    Identifiability of mixtures , author=. The annals of Mathematical statistics , volume=. 1961 , publisher=

    work page 1961

  53. [53]

    Wiley Interdisciplinary Reviews: Computational Statistics , volume=

    Unsupervised clustering using nonparametric finite mixture models , author=. Wiley Interdisciplinary Reviews: Computational Statistics , volume=. 2024 , publisher=

    work page 2024

  54. [54]

    Identifiability of nonparametric mixture models and bayes optimal clustering , author=

    work page

  55. [55]

    , author=

    IDENTIFIABILITY OF HIERARCHICAL LATENT ATTRIBUTE MODELS. , author=. Statistica Sinica , volume=

    work page

  56. [56]

    arXiv preprint arXiv:1502.06644 , year=

    On the identifiability of mixture models from grouped samples , author=. arXiv preprint arXiv:1502.06644 , year=

    work page arXiv

  57. [57]

    , author=

    Analysis of a complex of statistical variables into principal components. , author=. Journal of educational psychology , volume=. 1933 , publisher=

    work page 1933

  58. [58]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    Probabilistic principal component analysis , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 1999 , publisher=

    work page 1999

  59. [59]

    Signal processing , volume=

    Blind separation of sources, part I: An adaptive algorithm based on neuromimetic architecture , author=. Signal processing , volume=. 1991 , publisher=

    work page 1991

  60. [60]

    Signal processing , volume=

    Independent component analysis, a new concept? , author=. Signal processing , volume=. 1994 , publisher=

    work page 1994

  61. [61]

    1998 , publisher=

    Overview of factor analysis , author=. 1998 , publisher=

    work page 1998

  62. [62]

    IEEE transactions on pattern analysis and machine intelligence , volume=

    Convex and semi-nonnegative matrix factorizations , author=. IEEE transactions on pattern analysis and machine intelligence , volume=. 2008 , publisher=

    work page 2008

  63. [63]

    The pseudo-marginal approach for efficient Monte Carlo computations , author=

    work page

  64. [64]

    Neural computation , volume=

    Mixtures of probabilistic principal component analyzers , author=. Neural computation , volume=. 1999 , publisher=

    work page 1999

  65. [65]

    Genetics , volume=

    Estimation of population growth or decline in genetically monitored populations , author=. Genetics , volume=. 2003 , publisher=

    work page 2003

  66. [66]

    Computational Statistics & Data Analysis , volume=

    Modelling high-dimensional data by mixtures of factor analyzers , author=. Computational Statistics & Data Analysis , volume=. 2003 , publisher=

    work page 2003

  67. [67]

    2000 , publisher=

    Finite mixture models , author=. 2000 , publisher=

    work page 2000

  68. [68]

    Prevention Science , volume=

    Finite mixture models with student t distributions: an applied example , author=. Prevention Science , volume=. 2020 , publisher=

    work page 2020

  69. [69]

    Journal of computational and graphical statistics , volume=

    Mixtures of gamma distributions with applications , author=. Journal of computational and graphical statistics , volume=. 2001 , publisher=

    work page 2001

  70. [70]

    Statistica Sinica , pages=

    Finite mixture modelling using the skew normal distribution , author=. Statistica Sinica , pages=. 2007 , publisher=

    work page 2007

  71. [71]

    Semiparametric estimation of a two-component mixture model , author=

    work page

  72. [72]

    The Annals of Statistics , pages=

    Inference for mixtures of symmetric distributions , author=. The Annals of Statistics , pages=. 2007 , publisher=

    work page 2007

  73. [73]

    Econometric Theory , volume=

    Inference on two-component mixtures under tail restrictions , author=. Econometric Theory , volume=. 2017 , publisher=

    work page 2017

  74. [74]

    Technometrics , volume=

    Archetypal analysis , author=. Technometrics , volume=. 1994 , publisher=

    work page 1994

  75. [75]

    Journal of the American Mathematical Society , volume=

    Testing the manifold hypothesis , author=. Journal of the American Mathematical Society , volume=

    work page

  76. [76]

    Advances in Neural Information Processing Systems , volume=

    Consistent estimation of identifiable nonparametric mixture models from grouped observations , author=. Advances in Neural Information Processing Systems , volume=

    work page

  77. [77]

    Identifiability of parameters in latent structure models with many observed variables , author=

    work page

  78. [78]

    Nonparametric finite translation hidden Markov models and extensions , author=

    work page

  79. [79]

    The annals of statistics , volume=

    Nonparametric estimation of component distributions in a multivariate mixture , author=. The annals of statistics , volume=. 2003 , publisher=

    work page 2003

  80. [80]

    Annales de l'institut Fourier , volume=

    An application of classical invariant theory to identifiability in nonparametric mixtures , author=. Annales de l'institut Fourier , volume=

    work page

Showing first 80 references.