Pith. sign in

REVIEW 3 major objections 7 minor 114 references

Transfer Learning in High-Dimensional Clustering: Minimax Thresholds and Applications in Single-Cell Data

T0 review · 3 major / 7 minor · reviewed 2026-07-31 · grok-4.5

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

Pith's one-line read High-dimensional clustering succeeds by transfer only when source signal, alignment, and a product SNR all clear fixed scales; otherwise the target alone must clear the classical threshold.

desk verdict Clean high-d transfer phase diagram for GMM label recovery, nearly matching upper/lower bounds; log gap and isotropic-alignment model are the real limits. read the letter →

arxiv 2607.25031 v1 pith:CT4LEFZJ submitted 2026-07-27 math.ST stat.MEstat.MLstat.TH

Abhinav Chakraborty , Sagnik Nandy This is my paper · ORCID
classification math.STstat.MEstat.MLstat.TH MSC 62H3062C20
keywords transferlearninghigh-dimensionalclusteringGaussianmixturemodelsminimaxthresholdsspectralsingle-cellRNA-seqphasetransitionnegative
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 you must cluster a high-dimensional target sample that is too weak on its own, a related source sample can rescue the labels—but only if three scales all clear. The paper pins those scales for two-community Gaussian mixtures: the source must be strong enough to learn its own direction, the cosine alignment µ times target SNR must exceed a constant, and the product µ·source SNR·target SNR must beat the high-dimensional estimation cost √(d/n_S). Matching lower bounds show the same scales are necessary up to logs and fixed constants, so the phase diagram is essentially sharp. An adaptive selector picks the target-only route when the target already clears its classical threshold and the source route otherwise, at no extra asymptotic cost. The same geometry extends to multiple communities and multiple sources, and the procedures are competitive on a lung single-cell atlas.

What carries the argument

Project target observations onto an estimated source direction (singular vector when d ≲ n_S, label-then-average when d ≫ n_S) and take signs; consistency rests on a Gaussian/Haar anti-concentration argument that the projected means stay separable once the three transfer conditions hold.

What would settle it

In a synthetic two-community GMM with fixed d, n_T, n_S, drive µ·∆_S·∆_T just below √(d/n_S) while keeping the other two transfer conditions and the target below its classical threshold; any method’s misclustering rate must stay bounded away from zero.

Watch

Extended reading notes

Core claim

Consistent recovery of target labels in a high-d two-community GMM is possible if and only if either the target SNR alone exceeds the classical threshold max{1,(d/n_T)^{1/4}}, or the source is strong enough, well enough aligned, and the product µ·∆_S·∆_T clears √(d/n_S); the paper supplies a spectral procedure attaining the upper side and an Assouad lower bound matching the same scales up to logs.

Load-bearing premise

Relatedness is entirely captured by the cosine of the angle between isotropic Gaussian cluster-mean directions; if real sources differ mainly by non-directional batch effects or label-dependent noise, the stated thresholds need not govern transfer.

Editorial extensions

If this is right

  • Transfer is useless below the product scale even if the source is perfectly aligned and infinitely strong.
  • An adaptive validation statistic can switch routes without paying an extra SNR cost.
  • Pooling several sources works asymptotically whenever at least one source meets the alignment and SNR conditions.
  • The same geometric thresholds guide when reference atlases help label-transfer in scRNA-seq.

Reading between the lines

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

  • The remaining log gap is likely an artifact of exact source-label recovery; a softer source estimator may close it.
  • The same product-scale obstruction should appear in other high-d unsupervised transfer problems that first estimate a source direction then project.
  • Negative transfer observed when alignment is poor is predicted by the phase diagram and can be used as a diagnostic for batch mismatch.
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 / 7 minor

Summary. The paper studies transfer-assisted clustering in a high-dimensional two-community Gaussian mixture model, where one target and one (or several) source datasets share aligned but non-identical cluster-mean directions, with alignment measured by cosine similarity µ. The main results are: (i) a spectral meta-procedure (Algorithm 1) achieving consistent target clustering when either the target SNR satisfies ∆_T ≫ max{1,(d/n_T)^{1/4}} or the source satisfies ∆_S ≫ (d(log n_S)²/n_S)^{1/4}, µ∆_T ≫ 1, µ∆_S∆_T ≫ √(d/n_S) (Theorem 1); (ii) a matching necessary condition at fixed universal constants on the same four scales (Theorem 4), proved via an Assouad reduction on pairwise label products with four total-variation routes (target-revealed, source-revealed, doubly subcritical, product-scale) and a conditioning argument transferring a surrogate random-direction prior onto the deterministic alignment class; (iii) a validation-statistic-based adaptive selector that chooses the target or source branch without rate loss (Theorem 2); and (iv) extensions to K communities and m sources with subspace projection (Theorems 5–6). Simulations and a leave-one-batch-out analysis of a human lung scRNA-seq atlas accompany the theory.

Significance. To my knowledge this is the first minimax characterization of transfer learning for clustering in the genuinely high-dimensional regime (d possibly ≫ n), a setting the closest prior work (Tian et al. 2026) does not cover. The upper and lower bounds identify the same four scales — target, source, alignment, and the product scale µ∆_S∆_T ~ √(d/n_S) — which is a substantive and falsifiable characterization of when borrowing helps. The lower-bound construction (four TV routes plus the exact-alignment conditioning in Lemma S21/S22) and the Gaussian anti-concentration argument in the upper bound are techniques of independent interest. The authors disclose the remaining logarithmic gap and attribute it concretely to exact source-label recovery via Ndaoud's Theorem 8, which is appropriately transparent. The adaptive procedure and the multi-community/multi-source extensions broaden applicability, and the scRNA-seq analysis, while mixed (see major comment 1), is an honest empirical stress test including negative transfer. If the results hold, the paper provides a useful theoretical foundation for reference-based annotation pipelines such as label transfer.

major comments (3)
  1. [§6, Table 1] The real-data results do not demonstrate transfer benefit for any theoretically analyzed procedure: the adaptive estimator coincides with target-only on all four batches (identical ARI/V-measure/L_mult rows), and the multi-source pooled variant is strictly worse than target-only everywhere. The only method that improves on target-only (ASK440, ASK454) is the pooled estimator of Algorithm S1, for which §A.3 explicitly states no theoretical analysis is pursued. Given n_T ≈ 2–3k, d = 5000, K = 13, the data plausibly lie in the target-only regime of Theorem 5, so the adaptive rule is behaving as designed but the section cannot support the abstract's claim of 'practical effectiveness' of transfer. Please (a) state this regime diagnosis explicitly, ideally by checking estimated target SNR against (39); (b) add a target-subsampling experiment in the style of Figure 1 that evaluates Algorithm 3
  2. [§3, Theorem 4] The necessary conditions are fixed-constant requirements (∆_T ≥ c_1 max{1,(d/n_T)^{1/4}}, µ∆_T ≥ c_3, etc.), whereas the sufficient conditions in Theorem 1 require divergence. In the regime d = O(n_T), condition (A) of Theorem 4 only forces ∆_T ≥ c_1, but consistency is in fact impossible at bounded ∆_T (the minimax misclustering proportion in the two-component GMM is of order exp(-∆²/8) even with known direction; cf. Ndaoud 2022), so the lower bound leaves a divergence-versus-constant gap in addition to the disclosed logarithmic one. The abstract and §1.1 phrase 'characterize, up to logarithmic factors, the phase transition' should be qualified accordingly, or Theorem 4 strengthened to rule out bounded ∆_T (resp. bounded µ∆_T) when d/n_T (resp. the relevant aspect ratio) does not diverge. A short discussion of whether the boundary scales are achievable at large constants would also help
  3. [§2.3, Theorem 2 (and §4, Theorem 6)] The adaptivity guarantee is stated as a dichotomy between ∆_T ≫ max{1,(d/n_T)^{1/4}} (Case 1) and ∆_T ≤ D_0(1+(d/n_T)^{1/4}) for a fixed D_0 (Case 2), with the threshold constant C_0 in (18) chosen as a function of the unknown D_0. Two caveats are not surfaced in the main text: (i) the boundary regime ∆_T of the same order as (or slowly diverging relative to) the threshold is covered by neither case, so the claim in §1.1 that the adaptive procedure 'attains the same asymptotic success regimes' holds only on the union of the two covered regimes; (ii) C_0 is regime-dependent through D_0, so 'no additional cost for adaptation' is up to a constant that must be calibrated (the bootstrap heuristic of §A.2 is a reasonable but unanalyzed fix; α is a user tuning parameter). Please state the quantifiers precisely in Theorem 2 (and analogously for Theorem 6 under condition (50)) and temper the info
minor comments (7)
  1. [Throughout] Typos/grammar: 'and and' (abstract); 'transfer-assited' (§1, contribution 1); double comma 'sample size,,' (§1); 'An numerical experiment' (§5.1); 'the we use pZ_S' (before (49)); 'a a deep embedded clustering' (§6); 'realtively' (§A.1); garbled phrase 'spectral procedures depending on target, and source datasets' (Figure 1 caption); 'William M.' appears as a garbled author in the Stuart et al. reference.
  2. [Figure 2] §1.1 refers to yellow/green/blue/grey regions in Figure 2, but the caption does not define the color scheme; please make the figure self-contained and verify the colors render as intended.
  3. [§A.4, Algorithm S2 vs. Algorithm 3] The clustering module is called TSClust in Algorithm 3 and the text, but Algorithm S2 is titled 'TClust'; please make the naming consistent. Also confirm whether the module requires T_0 = 2 log n_T (Algorithm 3) or 3 log n (Theorem S4).
  4. [§4, Statistical Model] The factor-of-2 discrepancy between the two-community SNR definition (6) and the multi-community minimum-pairwise-separation definition in §4 is disclosed, but since Theorem 1 is quoted as a special case it would help to state the exact constant mapping once, rather than only 'up to a factor of 2'.
  5. [Table 1] The NMF baseline attains the best L_mult on ASK454 with ARI = 0 due to a degenerate partition; the caption's 'best value in bold' convention will therefore highlight a meaningless result. Please add a caption caveat, and consider whether the adaptive/multi-source rows being identical to target-only on all batches should be remarked upon in the caption as well.
  6. [§3, Eq. (24)] In §3, the reduction assumes n_T = 2k with the odd case handled by discarding one observation; please state explicitly in Lemma 1 / (24) that this changes nothing asymptotically, and note k/(2n_T) = 1/4 uses the even case.
  7. [§5 and §A.2] Simulations: the number of bootstrap replications B varies across experiments (50 in §5.1, 20 in §A.1, 10 in §5.2, 30 in §6); a brief sensitivity note on B and α would strengthen reproducibility. Reporting runtime of the adaptive and pooled procedures relative to TL-GMM/TGMM would also be useful.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: phase transition is derived from independent upper/lower bound arguments, not from fitted inputs renamed as predictions.

full rationale

The paper’s central claim is a minimax phase transition for consistent target label recovery in a high-dimensional two-community GMM with directional alignment. The risk Rp(∆T, ∆S, µ) is defined externally via the misclustering loss on the parameter class Ω; sufficiency (Theorem 1) comes from a spectral/source-projection procedure with concentration and anti-concentration analysis; necessity (Theorem 4) comes from an Assouad reduction to neighboring mixture laws and four total-variation routes. Neither side fits a parameter to the same quantity it then “predicts.” Adaptive selection uses a validation statistic with uniform concentration (Theorem 2); the parametric bootstrap only chooses the absolute constant C0/D0 in practice and is not load-bearing for the asymptotic theorems. Citations to Ndaoud, Löffler et al., Giraud–Verzelen, and Assouad/Yu are external clustering and lower-bound tools. Self-citations (e.g. Chakraborty–Maity on multi-source adaptation cost) are peripheral. No self-definitional loop, fitted-input-as-prediction, or uniqueness-imported-from-authors structure appears in the derivation chain.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The central phase transition rests on isotropic two- (or multi-) community Gaussian mixture models, deterministic labels, cosine alignment of means, and standard high-d spectral/Assouad machinery. Free parameters are only practical tuning constants for adaptive thresholds, not loads on the asymptotic claim. No new physical entities are invented.

free parameters (3)
  • C_0 (two-community validation threshold multiplier) = bootstrap quantile; α=0.5 in experiments
    Absolute constant in τ_n; theory only needs some large enough C_0. In practice chosen by parametric bootstrap with user α (default 0.5).
  • D_0 (multi-community validation threshold multiplier) = bootstrap quantile; α=0.5
    Analogous absolute constant in t_n for K>2; bootstrap-calibrated in simulations and lung analysis.
  • TGMM/TL-GMM comparator hyperparameters (λ, κ, C_λ0) = λ=0.5; κ=1/3; C_λ0=1.7 or 5-fold CV
    Affect only benchmark comparisons, not the paper’s own theorems; still chosen by hand/CV in experiments.
assumptions (6)
  • domain assumption Target and source observations are isotropic Gaussian mixtures with deterministic ±1 (or one-hot) labels and independent noise across datasets.
    Model (3)–(4) and multi-cluster (32)–(33); standard but strong for scRNA-seq after log-normalization.
  • domain assumption Source–target relatedness is lower-bounded cosine similarity µ of cluster-mean directions (or pairwise contrasts).
    Equation (5) and multi-cluster (34); defines the parameter class Ω.
  • standard math High-dimensional spectral clustering / relaxed K-means achieve exact or consistent recovery at the known minimax SNR scales (Ndaoud; Giraud–Verzelen; Gao–Zhang).
    Used as GOOD-CLUSTERER and in multi-cluster initialization; cited theorems imported as black boxes.
  • standard math Assouad’s lemma and total-variation / Hellinger calculus for Gaussian mixture experiments control minimax misclustering risk.
    Section 3 reduction via pairwise products τ_j.
  • domain assumption Cluster sizes are balanced up to constant β; for multi-cluster source, informative source mean matrix has full rank K and σ_K(Θ) ≳ ∆_S σ_S.
    Conditions (37)–(38) and (35); authors note full rank is slightly stronger than necessary.
  • ad hoc to paper For adaptive multi-cluster selection, all pairwise target separations are comparable up to constant κ (condition (50)).
    Needed so the validation statistic does not accept target-only when only easy contrasts are strong; Remark 6 admits this.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Transfer Learning in High-Dimensional Clustering: Minimax Thresholds and Applications in Single-Cell Data." pith.science (2026). https://pith.science/paper/CT4LEFZJ

@misc{pith2026260725031,
  author       = {Pith},
  title        = {Pith review of: Transfer Learning in High-Dimensional Clustering: Minimax Thresholds and Applications in Single-Cell Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CT4LEFZJ}},
  note         = {Machine review of arXiv:2607.25031}
}
read the original abstract

Clustering is a fundamental problem in statistics, with applications across many scientific disciplines. In many modern applications involving clustering, the primary dataset (the target data) is accompanied by related datasets (the source data). Transferring information from such sources may improve clustering accuracy in the target, making transfer learning for clustering practically important. Despite recent progress, the conditions under which source data improve target clustering remain unclear in high-dimensional settings, even for the canonical Gaussian mixture model. In this paper, we study the clustering problem in a two-community Gaussian mixture model where relatedness is captured by the geometric alignment of the target and source cluster means. We develop a minimax-optimal transfer-assisted clustering procedure and characterize, up to logarithmic factors, the phase transition for consistent target clustering in terms of the signal-to-noise ratios, sample sizes, ambient dimension, and degree of alignment between the datasets. The technique is also extended to adaptively choose between the target-only or the source assisted clustering depending on the target signal strength. Furthermore, we also extend our techniques to accommodate multiple communities and and multiple source datasets. Extensive simulations and an analysis of a human lung single-cell RNA-sequencing atlas demonstrate the practical effectiveness of our methods.

Figures

Figures reproduced from arXiv: 2607.25031 by the authors.

Figure 1
Figure 1. Transfer-assisted clustering on the human lung atlas data from [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Schematic phase diagram in the two-community, one-source setting, shown as a function of the [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Target misclustering error as a function of the alignment parameter [PITH_FULL_IMAGE:figures/full_fig_p019_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Overall and pairwise target misclustering errors for the target-only, source-1-only, source-2-only, [PITH_FULL_IMAGE:figures/full_fig_p020_4.png]
Figure 5
Figure 5. Figure 5: Target misclustering error as a function of the alignment parameter [PITH_FULL_IMAGE:figures/full_fig_p021_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

114 extracted references · 2 canonical work pages

  1. [1]

    Algorithms in bioinformatics:

    Jointly embedding multiple single-cell omics measurements , author=. Algorithms in bioinformatics:... International Workshop, WABI..., proceedings. WABI (Workshop) , volume=

  2. [2]

    IEEE Transactions on Information Theory , volume=

    Bias-corrected joint spectral embedding for multilayer networks with invariant subspace: entrywise eigenvector perturbation and inference , author=. IEEE Transactions on Information Theory , volume=. 2024 , publisher=

    2024

  3. [3]

    Zhou , title =

    Yu Lu and Harrison H. Zhou , title =. arXiv preprint arXiv:1612.02099 , year =

    arXiv

  4. [4]

    Journal of Computer and System Sciences , volume=

    A spectral algorithm for learning mixture models , author=. Journal of Computer and System Sciences , volume=. 2004 , publisher=

    2004

  5. [5]

    Conference On Learning Theory , pages=

    Hidden integrality of SDP relaxations for sub-Gaussian mixture models , author=. Conference On Learning Theory , pages=. 2018 , organization=

    2018

  6. [6]

    Statistics and computing , volume=

    A tutorial on spectral clustering , author=. Statistics and computing , volume=. 2007 , publisher=

    2007

  7. [7]

    IEEE Transactions on Information Theory , volume=

    Optimal estimation of shared singular subspaces across multiple noisy matrices , author=. IEEE Transactions on Information Theory , volume=

  8. [8]

    arXiv preprint arXiv:2507.22170 , year=

    Stacked SVD or SVD stacked? A Random Matrix Theory perspective on data integration , author=. arXiv preprint arXiv:2507.22170 , year=

    arXiv

Show all 114 references
  1. [9]

    arXiv preprint arXiv:2605.09471 , year=

    The Statistical Cost of Adaptation in Multi-Source Transfer Learning , author=. arXiv preprint arXiv:2605.09471 , year=

    arXiv

  2. [10]

    The Annals of Statistics , volume=

    Global and individualized community detection in inhomogeneous multilayer networks , author=. The Annals of Statistics , volume=. 2022 , publisher=

    2022

  3. [11]

    IEEE/CAA Journal of Automatica Sinica , volume=

    A survey on negative transfer , author=. IEEE/CAA Journal of Automatica Sinica , volume=. 2022 , publisher=

    2022

  4. [12]

    IEEE Transactions on knowledge and data engineering , volume=

    A survey on transfer learning , author=. IEEE Transactions on knowledge and data engineering , volume=. 2009 , publisher=

    2009

  5. [13]

    Journal of Big Data , volume=

    A survey of transfer learning , author=. Journal of Big Data , volume=

  6. [14]

    Machine Learning , volume=

    A theory of learning from different domains , author=. Machine Learning , volume=

  7. [15]

    Monthly Notices of the Royal Astronomical Society , volume =

    Chattopadhyay, Souradeep and Maitra, Ranjan , title =. Monthly Notices of the Royal Astronomical Society , volume =. 2017 , month =

    2017

  8. [16]

    Robust Discovery of Mild Cognitive Impairment Subtypes and Their Risk of Alzheimer's Disease Conversion Using Unsupervised Machine Learning and Gaussian Mixture Modeling

    Nezhadmoghadam, Fahimeh and Martinez-Torteya, Antonio and Trevi \ n o, Victor and Mart \' nez, Emmanuel and Santos, Alejandro and Tamez-Pe \ n a, Jose and Alzheimer's Disease Neuroimaging Initiative. Robust Discovery of Mild Cognitive Impairment Subtypes and Their Risk of Alzh...

  9. [17]

    The Annals of Statistics , volume=

    Fundamental limits of community detection from multi-view data: multi-layer, dynamic and partially labeled block models , author=. The Annals of Statistics , volume=. 2025 , publisher=

    2025

  10. [18]

    arXiv preprint arXiv:2605.19894 , year=

    Sharp Spectral Thresholds for Multi-View Spiked Wigner Models , author=. arXiv preprint arXiv:2605.19894 , year=

    arXiv

  11. [19]

    Zhang , title =

    Chao Gao and Anderson Y. Zhang , title =. The Annals of Statistics , volume =

  12. [20]

    Tony and Zhang, Anru , title =

    Cai, T. Tony and Zhang, Anru , title =. The Annals of Statistics , volume =

  13. [21]

    Achieving Optimal Clustering in Gaussian Mixture Models with Anisotropic Covariance Structures , year =

    Chen, Xin and Zhang, Anderson Ye , booktitle =. Achieving Optimal Clustering in Gaussian Mixture Models with Anisotropic Covariance Structures , year =

  14. [22]

    The Annals of Statistics , volume =

    Ndaoud, Mohamed , title =. The Annals of Statistics , volume =

  15. [23]

    Optimality of Spectral Clustering in the Gaussian Mixture Model , journal =

    L. Optimality of Spectral Clustering in the Gaussian Mixture Model , journal =

  16. [24]

    Ieee Access , volume=

    Medical image segmentation with limited supervision: a review of deep network models , author=. Ieee Access , volume=. 2021 , publisher=

    2021

  17. [25]

    Journal of Machine Learning Research , volume=

    Community detection and stochastic block models: recent developments , author=. Journal of Machine Learning Research , volume=

  18. [26]

    Cell , year =

    Comprehensive Integration of Single-Cell Data , author =. Cell , year =

  19. [27]

    and Ren, Xing and Guenthoer, Jamie and Smythe, Kimberly S

    Zhao, Edward and Stone, Matthew R. and Ren, Xing and Guenthoer, Jamie and Smythe, Kimberly S. and Pulliam, Thomas and Williams, Stephen R. and Uytingco, Cedric R. and Taylor, Sarah E. and Nghiem, Paul and Bielas, Jason H. and Gottardo, Raphael , journal =. Spatial transcriptom...

  20. [28]

    and Andrews, Tallulah and Yiu, Andrew and Chandra, Tamir and Natarajan, Kedar N

    Kiselev, Vladimir Yu and Kirschner, Kristina and Schaub, Michael T. and Andrews, Tallulah and Yiu, Andrew and Chandra, Tamir and Natarajan, Kedar N. and Reik, Wolf and Barahona, Mauricio and Green, Anthony R. and Hemberg, Martin , title=. Nature Methods , year=

  21. [29]

    Spatial based expectation maximizing ( EM )

    Balafar, M A. Spatial based expectation maximizing ( EM ). Diagn Pathol

  22. [30]

    Annals of statistics , pages=

    Adaptive estimation of a quadratic functional by model selection , author=. Annals of statistics , pages=. 2000 , publisher=

    2000

  23. [31]

    2025 , publisher=

    High-dimensional probability , author=. 2025 , publisher=

    2025

  24. [32]

    , author=

    Introduction to the non-asymptotic analysis of random matrices. , author=

  25. [33]

    Transfer Clustering Based on Gaussian Mixture Model , year=

    Wang, Rongrong and Zhou, Jin and Liu, Xiangdao and Han, Shiyuan and Wang, Lin and Chen, Yuehui , booktitle=. Transfer Clustering Based on Gaussian Mixture Model , year=

  26. [34]

    Journal of the Royal Statistical Society , volume=

    Contributions to the mathematical theory of evolution , author=. Journal of the Royal Statistical Society , volume=. 1893 , publisher=

  27. [35]

    Journal of the royal statistical society: series B (methodological) , volume=

    Maximum likelihood from incomplete data via the EM algorithm , author=. Journal of the royal statistical society: series B (methodological) , volume=. 1977 , publisher=

    1977

  28. [36]

    , author=

    A Probabilistic Analysis of EM for Mixtures of Separated, Spherical Gaussians. , author=. Journal of Machine Learning Research , volume=

  29. [37]

    Advances in Neural Information Processing Systems , volume=

    Global analysis of expectation maximization for mixtures of two gaussians , author=. Advances in Neural Information Processing Systems , volume=

  30. [38]

    Conference on Learning Theory , pages=

    Ten steps of EM suffice for mixtures of two Gaussians , author=. Conference on Learning Theory , pages=. 2017 , organization=

    2017

  31. [39]

    Bulletin de l'Acad

    Steinhaus, Hugo , title =. Bulletin de l'Acad. 1956 , volume =

    1956

  32. [40]

    , title =

    Ward Jr., Joe H. , title =. Journal of the American Statistical Association , year =

  33. [41]

    , title =

    Forgy, Edward W. , title =. Biometrics , year =

  34. [42]

    , title =

    Lloyd, Stuart P. , title =. IEEE Transactions on Information Theory , year =

  35. [43]

    Nature methods , volume=

    Fast, sensitive and accurate integration of single-cell data with Harmony , author=. Nature methods , volume=. 2019 , publisher=

    2019

  36. [44]

    A cellular census of human lungs identifies novel cell states in health and in asthma

    Vieira Braga, Felipe A and Kar, Gozde and Berg, Marijn and Carpaij, Orestes A and Polanski, Krzysztof and Simon, Lukas M and Brouwer, Sharon and Gomes, Tom \'a s and Hesse, Laura and Jiang, Jian and Fasouli, Eirini S and Efremova, Mirjana and Vento-Tormo, Roser and Talavera-L ...

  37. [45]

    Zhang , title =

    Xin Chen and Anderson Y. Zhang , title =. Advances in Neural Information Processing Systems (NeurIPS) , year =

  38. [46]

    arXiv preprint arXiv:2111.07041 , year =

    Stanislav Minsker and Mohamed Ndaoud , title =. arXiv preprint arXiv:2111.07041 , year =

    arXiv

  39. [47]

    Universal Lower Bounds and Optimal Rates: Achieving Minimax Clustering Error in Sub-Exponential Mixture Models , booktitle =

    Maximilien Dreveton and Alperen G. Universal Lower Bounds and Optimal Rates: Achieving Minimax Clustering Error in Sub-Exponential Mixture Models , booktitle =. 2024 , note =

    2024

  40. [48]

    IEEE Transactions on Information Theory , year=

    Adversarially robust clustering with optimality guarantees , author=. IEEE Transactions on Information Theory , year=

  41. [49]

    arXiv preprint arXiv:2509.08237 , year =

    Xin Bing and Dehan Kong and Bingqing Li , title =. arXiv preprint arXiv:2509.08237 , year =

  42. [50]

    Proceedings of the 20th International Conference on Artificial Intelligence and Statistics (AISTATS) , series =

    Tianyang Li and Xinyang Yi and Constantine Caramanis and Pradeep Ravikumar , title =. Proceedings of the 20th International Conference on Artificial Intelligence and Statistics (AISTATS) , series =

  43. [51]

    The Annals of Statistics , volume =

    Florentina Bunea and Christophe Giraud and Xi Luo and Martin Royer and Nicolas Verzelen , title =. The Annals of Statistics , volume =. 2020 , doi =

    2020

  44. [52]

    Tsybakov , title =

    Mohamed Ndaoud and Alexandre B. Tsybakov , title =. IEEE Transactions on Information Theory , volume =. 2020 , doi =

    2020

  45. [53]

    Zhang and Harrison H

    Anderson Y. Zhang and Harrison H. Zhou , title =. The Annals of Statistics , volume =. 2016 , doi =

    2016

  46. [54]

    Statistical Science , volume =

    Chao Gao and Zongming Ma , title =. Statistical Science , volume =. 2021 , doi =

    2021

  47. [55]

    IEEE Transactions on Information Theory , volume =

    Yingjie Fei and Yudong Chen , title =. IEEE Transactions on Information Theory , volume =. 2020 , doi =

    2020

  48. [56]

    Zhou , title =

    Yihong Wu and Harrison H. Zhou , title =. arXiv preprint arXiv:1908.10935 , year =

    1908 arXiv

  49. [57]

    Wainwright and Bin Yu , title =

    Sivaraman Balakrishnan and Martin J. Wainwright and Bin Yu , title =. The Annals of Statistics , volume =. 2017 , doi =

    2017

  50. [58]

    Lloyd , title =

    Stuart P. Lloyd , title =. IEEE Transactions on Information Theory , volume =. 1982 , doi =

    1982

  51. [59]

    Bernoulli , volume =

    Florentina Bunea and Patrick Jreije and Claudiu Lefter and Martin Wegkamp , title =. Bernoulli , volume =. 2022 , doi =

    2022

  52. [60]

    Mathematical Statistics and Learning , volume =

    Christophe Giraud and Nicolas Verzelen , title =. Mathematical Statistics and Learning , volume =

  53. [61]

    arXiv preprint arXiv:2411.17054 , year =

    Zhengchi Ma and Rong Ma , title =. arXiv preprint arXiv:2411.17054 , year =

    arXiv

  54. [62]

    arXiv preprint arXiv:2501.09336 , year =

    Yuepeng Yang and Cong Ma , title =. arXiv preprint arXiv:2501.09336 , year =

    arXiv

  55. [63]

    arXiv preprint arXiv:2206.04306 , year =

    Runbing Zheng and Minh Tang , title =. arXiv preprint arXiv:2206.04306 , year =

  56. [64]

    IEEE Journal of Biomedical and Health Informatics , volume =

    Zile Wang and Fengyu Lei and Xiaoping Shi and Jianping Zhao and Junfeng Xia , title =. IEEE Journal of Biomedical and Health Informatics , volume =. 2025 , doi =

    2025

  57. [65]

    Journal of the American Statistical Association , year =

    Yuqi Gu and Zhongyuan Lyu and Kaizheng Wang , title =. Journal of the American Statistical Association , year =

  58. [66]

    Pattern Recognition , volume =

    Zhihao Zhang and Jie Zhou , title =. Pattern Recognition , volume =. 2012 , doi =

    2012

  59. [67]

    IEEE Transactions on Cybernetics , volume =

    Rongrong Wang and Shiyuan Han and Jin Zhou and Yuehui Chen and Lin Wang and Tao Du and Ke Ji and Ya-Ou Zhao and Kun Zhang , title =. IEEE Transactions on Cybernetics , volume =

  60. [68]

    Journal of the American Statistical Association , pages=

    Robust unsupervised multi-task and transfer learning on gaussian mixture models , author=. Journal of the American Statistical Association , pages=

  61. [69]

    Lock and Katherine A

    Eric F. Lock and Katherine A. Hoadley and J. S. Marron and Andrew B. Nobel , title =. The Annals of Applied Statistics , volume =. 2013 , doi =

    2013

  62. [70]

    Qing Feng and Meilei Jiang and Jan Hannig and J. S. Marron , title =. Journal of Multivariate Analysis , volume =. 2018 , doi =

    2018

  63. [71]

    Tony Cai and Anru Zhang , title =

    T. Tony Cai and Anru Zhang , title =. The Annals of Statistics , volume =. 2018 , doi =

    2018

  64. [72]

    arXiv preprint arXiv:1812.08078 , year =

    Mohamed Ndaoud , title =. arXiv preprint arXiv:1812.08078 , year =

    arXiv

  65. [73]

    Statistical Science , volume =

    Chao Gao and Zongming Ma , title =. Statistical Science , volume =

  66. [74]

    Welch and Velina Kozareva and Ashley Ferreira and Charles Vanderburg and Carly Martin and Evan Z

    Joshua D. Welch and Velina Kozareva and Ashley Ferreira and Charles Vanderburg and Carly Martin and Evan Z. Macosko , title =. Cell , volume =. 2019 , doi =

    2019

  67. [75]

    Journal of the American statistical association , volume=

    Probability inequalities for sums of bounded random variables , author=. Journal of the American statistical association , volume=. 1963 , publisher=

    1963

  68. [76]

    Nature Biotechnology , volume =

    Zhi-Jie Cao and Ge Gao , title =. Nature Biotechnology , volume =. 2022 , doi =

    2022

  69. [77]

    A cellular census of human lungs identifies novel cell states in health and in asthma

    Vieira Braga, F A and Kar, G and Berg, M and Carpaij, O A and Polanski, K and Simon, L M and Brouwer, S and Gomes, T and Hesse, L and Jiang, J and Fasouli, E S and Efremova, M and Vento-Tormo, R and Talavera-L \'o pez, C and Jonker, M R and Affleck, K and Palit, S and Strzelec...

  70. [78]

    Bi-order Multimodal Integration of Single-cell Data , journal =

    Jinzhuang Dou and Shaoheng Liang and Vakul Mohanty and Qi Miao and Yuefan Huang and Qingnan Liang and Xuesen Cheng and Sangbae Kim and Jongsu Choi and Yumei Li and Li Li and May Daher and Rafet Ba. Bi-order Multimodal Integration of Single-cell Data , journal =. 2022 , doi =

    2022

  71. [79]

    Gromov--

    Facundo M. Gromov--. Foundations of Computational Mathematics , volume =. 2011 , doi =

    2011

  72. [80]

    Gromov--

    Gabriel Peyr. Gromov--. Proceedings of the 33rd International Conference on Machine Learning (ICML) , series =. 2016 , note =

    2016

  73. [81]

    Journal of Computational Biology , volume =

    Pinar Demetci and Rebecca Santorella and Bjorn Sandstede and William Stafford Noble and Ritambhara Singh , title =. Journal of Computational Biology , volume =. 2022 , doi =

    2022

  74. [82]

    Mauck and Yuhan Hao and Marlon Stoeckius and Peter Smibert and Rahul Satija , title =

    Tim Stuart and Andrew Butler and Paul Hoffman and Christoph Hafemeister and Efthymia Papalexi and William M. Mauck and Yuhan Hao and Marlon Stoeckius and Peter Smibert and Rahul Satija , title =. Cell , volume =

  75. [83]

    Mauck and Shiwei Zheng and Andrew Butler and Maddie J

    Yuhan Hao and Stephanie Hao and Erica Andersen-Nissen and William M. Mauck and Shiwei Zheng and Andrew Butler and Maddie J. Lee and Aaron J. Wilk and Charlotte Darby and Michael Zagar and Paul Hoffman and Marlon Stoeckius and Efthymia Papalexi and Eleni P. Mimitou and Jaison J...

    2021

  76. [84]

    Biometrics , volume =

    Liu, Tianqi and Lu, Yu and Zhu, Biqing and Zhao, Hongyu , title =. Biometrics , volume =. 2023 , month =. doi:10.1111/biom.13665 , url =

    2023 doi

  77. [85]

    ACM SIGMoD record , volume=

    Fast algorithms for projected clustering , author=. ACM SIGMoD record , volume=. 1999 , publisher=

    1999

  78. [86]

    Proceedings of the 1998 ACM SIGMOD international conference on Management of data , pages=

    Automatic subspace clustering of high dimensional data for data mining applications , author=. Proceedings of the 1998 ACM SIGMOD international conference on Management of data , pages=

    1998

  79. [87]

    Clustering high-dimensional data: A survey on subspace clustering, pattern-based clustering, and correlation clustering , year =

    Kriegel, Hans-Peter and Kr\". Clustering high-dimensional data: A survey on subspace clustering, pattern-based clustering, and correlation clustering , year =. ACM Trans. Knowl. Discov. Data , month = mar, articleno =. doi:10.1145/1497577.1497578 , abstract =

  80. [88]

    Computational Statistics & Data Analysis , author=

    High-dimensional data clustering , year=. Computational Statistics & Data Analysis , author=. doi:None , url=

  81. [89]

    Journal of the American Statistical Association , volume=

    Variable selection for model-based clustering , author=. Journal of the American Statistical Association , volume=. 2006 , publisher=

    2006

  82. [90]

    Journal of classification , volume=

    Comparing partitions , author=. Journal of classification , volume=

  83. [91]

    Xavier and Alaerts, Kaat and Anderson, Jeffrey S

    Di Martino, Adriana and Yan, Chao-Gan and Li, Qingyang and Denio, Erin and Castellanos, F. Xavier and Alaerts, Kaat and Anderson, Jeffrey S. and Assaf, Michal and Bookheimer, Susan Y. and Dapretto, Mirella and Deen, Ben and Delmonte, Sonja and Dinstein, Ilan and Ertl-Wagner, B...

    2014

  84. [92]

    V -Measure: A Conditional Entropy-Based External Cluster Evaluation Measure

    Rosenberg, Andrew and Hirschberg, Julia. V -Measure: A Conditional Entropy-Based External Cluster Evaluation Measure. Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning ( EMNLP - C o NLL ). 2007

    2007

  85. [93]

    Mieth, Bettina and Hockley, James R. F. and G. Using transfer learning from prior reference knowledge to improve the clustering of single-cell RNA-Seq data , journal=. 2019 , month=

    2019

  86. [94]

    arXiv preprint arXiv:2512.09259 , year=

    MoDaH achieves rate optimal batch correction , author=. arXiv preprint arXiv:2512.09259 , year=

  87. [95]

    Briefings in Bioinformatics , volume =

    Wang, Yu Mei and Sun, Yuzhi and Wang, Beiying and Wu, Zhiping and He, Xiao Ying and Zhao, Yuansong , title =. Briefings in Bioinformatics , volume =. 2024 , month =

    2024

  88. [96]

    Comptes rendus des s

    Deux remarques sur l'estimation , author=. Comptes rendus des s

  89. [97]

    Assouad, Fano, and Le Cam

    Yu, Bin. Assouad, Fano, and Le Cam. Festschrift for Lucien Le Cam: Research Papers in Probability and Statistics. 1997

    1997

  90. [98]

    Biometrika , volume=

    A useful variant of the Davis--Kahan theorem for statisticians , author=. Biometrika , volume=. 2015 , publisher=

    2015

  91. [99]

    Journal of the American Statistical Association , volume=

    A framework for feature selection in clustering , author=. Journal of the American Statistical Association , volume=. 2010 , publisher=

    2010

  92. [100]

    Soda , volume=

    k-means++: The advantages of careful seeding , author=. Soda , volume=

  93. [101]

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

    Empirical Bayes PCA in high dimensions , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 2022 , publisher=

    2022

  94. [102]

    Journal of the American Statistical Association , volume=

    Statistical significance of clustering for high-dimension, low--sample size data , author=. Journal of the American Statistical Association , volume=. 2008 , publisher=

    2008

  95. [103]

    Advances in Neural Information Processing Systems , volume=

    Minimax theory for high-dimensional gaussian mixtures with sparse mean separation , author=. Advances in Neural Information Processing Systems , volume=

  96. [104]

    CHIME: Clustering of high-dimensional Gaussian mixtures with EM algorithm and its optimality , volume =

    Cai, T and Ma, Jing and Zhang, Linjun , year =. CHIME: Clustering of high-dimensional Gaussian mixtures with EM algorithm and its optimality , volume =. Annals of Statistics , doi =

  97. [105]

    Influential features

    Jin, Jiashun and Wang, Wanjie , journal =. Influential features. 2016 , month =

    2016

  98. [106]

    Cell , volume =

    Droplet Barcoding for Single-Cell Transcriptomics Applied to Embryonic Stem Cells , author =. Cell , volume =. 2015 , month = may, doi =

    2015

  99. [107]

    Molecular cell , volume=

    Comparative analysis of single-cell RNA sequencing methods , author=. Molecular cell , volume=. 2017 , publisher=

    2017

  100. [108]

    CEL-Seq: Single-Cell RNA-Seq by Multiplexed Linear Amplification , journal =

    Tamar Hashimshony and Florian Wagner and Noa Sher and Itai Yanai , abstract =. CEL-Seq: Single-Cell RNA-Seq by Multiplexed Linear Amplification , journal =. 2012 , issn =. doi:https://doi.org/10.1016/j.celrep.2012.08.003 , url =

    2012 doi

  101. [109]

    and Rozenblatt-Rosen, Orit and Dor, Yuval and Regev, Aviv and Yanai, Itai , journal =

    Hashimshony, Tamar and Senderovich, Naftalie and Avital, Gal and Klochendler, Agnes and de Leeuw, Yaron and Anavy, Leon and Gennert, Dave and Li, Shuqiang and Livak, Kenneth J. and Rozenblatt-Rosen, Orit and Dor, Yuval and Regev, Aviv and Yanai, Itai , journal =. 2016 , month ...

    2016

  102. [110]

    Psychometrika , volume =

    Eckart, Carl and Young, Gale , title =. Psychometrika , volume =

  103. [111]

    , title =

    Franklin, Joel N. , title =. 1993 , isbn =

    1993

  104. [112]

    2019 , publisher =

    High-Dimensional Statistics: A Non-Asymptotic Viewpoint , author =. 2019 , publisher =

    2019

  105. [113]

    , journal =

    DeLaughter, Daniel M. , journal =. The Use of the. 2018 , month = apr, doi =

    2018

  106. [114]

    Picelli, Simone and Faridani, Omid R. and Bj. Full-Length. Nature Protocols , volume =. 2014 , month = jan, doi =

    2014

Pith tools

Reviewed July 31, 2026 · model on record in the stance chip above.