arxiv: 2605.00363 · v1 · submitted 2026-05-01 · 🧮 math.ST · stat.TH

Recognition: unknown

Profile Likelihood Inference for Anisotropic Hyperbolic Wrapped Normal Models on Hyperbolic Space

Kisung You

Pith reviewed 2026-05-09 19:01 UTC · model grok-4.3

classification 🧮 math.ST stat.TH

keywords hyperbolic spacewrapped normal distributionprofile likelihoodasymptotic normalityminimax estimationmanifold-valued dataconsistency

0 comments

The pith

The profile maximum likelihood estimator for the location in anisotropic hyperbolic wrapped normal models attains the Hájek-Le Cam local asymptotic minimax lower bound under squared geodesic loss.

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

This paper establishes profile likelihood inference for the location parameter of an anisotropic hyperbolic wrapped normal distribution on hyperbolic space. The covariance is treated as a nuisance parameter profiled out of the likelihood, with the estimator constrained to a covariance shell for finite-sample existence. The work proves strong consistency, joint asymptotic normality, and that the estimator achieves the theoretical minimax bound for estimating the location under geodesic loss, which matters for reliable inference on data with hyperbolic geometry such as hierarchies or networks.

Core claim

For independent observations from the anisotropic hyperbolic wrapped normal family, the profile maximum likelihood estimator for the manifold-valued location parameter, obtained by optimizing the likelihood after profiling out the covariance on a shell that bounds eigenvalues away from zero and infinity, is strongly consistent when the true covariance is interior to the shell. In global normal coordinates for location and log-covariance coordinates for the nuisance, the estimator is jointly asymptotically normal, and efficient inference for the location follows from the Schur-complement information matrix. The profile estimator attains the Hájek-Le Cam local asymptotic minimax lower bound

What carries the argument

The profile maximum likelihood estimator for the location parameter, constructed by maximizing the profiled likelihood over the manifold location while constraining the covariance to the eigenvalue-bounded shell.

If this is right

The constrained likelihood is well-posed and the estimator exists for finite samples.
The family is identifiable and the estimator is strongly consistent under the interior covariance condition.
Joint asymptotic normality holds for the location and log-covariance parameters.
Efficient profile inference for the location is available through the Schur complement of the information matrix.
The estimator attains the local asymptotic minimax lower bound under the stated loss conditions.

Where Pith is reading between the lines

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

Such profile methods may generalize to inference for location parameters in other distributions on Riemannian manifolds.
Attainment of the minimax bound suggests this estimator is optimal for recovering positions in hyperbolic embeddings of hierarchical data.
The explicit spectral clipping computation could enable scalable implementations for large datasets on hyperbolic space.

Load-bearing premise

The true covariance lies in the interior of the shell bounding eigenvalues away from zero and infinity, and the observations are independent and identically distributed from the anisotropic hyperbolic wrapped normal family.

What would settle it

A Monte Carlo experiment generating data from the model with covariance approaching the shell boundary and showing that the scaled risk of the profile estimator exceeds the asymptotic minimax bound would falsify the attainment result.

Figures

Figures reproduced from arXiv: 2605.00363 by Kisung You.

**Figure 1.** Figure 1: Scaled location risk n E{ρ(ˆµn, µ0) 2} compared with the efficient asymptotic target tr(I −1 αα·β ). normality of the anisotropic HWN experiment and derive the corresponding Hajek-Le Cam local asymptotic minimax lower bound under squared geodesic loss. The profile estimator attains the bound for truncated squared geodesic loss, and for ordinary squared geodesic loss under a stated local uniform-integrabili… view at source ↗

**Figure 2.** Figure 2: Empirical coverage of nominal 95% Wald regions. The location coverage uses the Schur view at source ↗

read the original abstract

We study likelihood-based inference for the anisotropic hyperbolic wrapped normal distribution on standard hyperbolic space. The model has a manifold-valued location parameter and a full positive definite covariance matrix in tangent coordinates. For independent observations from this family, we analyze the profile maximum likelihood estimator obtained by optimizing the likelihood over the location after profiling out the covariance. To guarantee finite-sample existence, we formulate the estimator on a covariance shell that bounds eigenvalues away from zero and infinity. We prove that this constrained likelihood is well posed, that the anisotropic wrapped normal family is identifiable, and that the estimator is strongly consistent when the true covariance lies in the interior of the shell. In global normal coordinates for the location and log-covariance coordinates for the nuisance parameter, we establish joint asymptotic normality and derive efficient profile inference for the location parameter through the Schur-complement information. We further prove local asymptotic normality of the experiment and obtain the H\'ajek--Le Cam local asymptotic minimax lower bound under squared geodesic loss. The profile estimator attains this bound for truncated squared loss, and for ordinary squared loss under a uniform-integrability condition. We also give an explicit computational form of the estimator based on spectral clipping of the empirical tangent covariance, and present a Monte Carlo calibration study showing that the finite-sample scaled location risk and Wald coverage agree with the asymptotic theory.

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.

Desk Editor's Note private letter to a colleague

The paper gives a complete asymptotic analysis for profile likelihood in the anisotropic hyperbolic wrapped normal but the uniform integrability for the minimax result under squared geodesic loss looks like it needs extra work.

read the letter

The main takeaway is that this work supplies new consistency, joint asymptotic normality, and local asymptotic minimax results for the profile MLE in the anisotropic hyperbolic wrapped normal on hyperbolic space, extending prior Euclidean or isotropic cases with a full covariance in tangent coordinates. They constrain the covariance to a shell for finite-sample existence, prove identifiability and strong consistency when the true parameter is interior, and derive efficient inference for the location via the Schur complement after profiling. The LAN argument and attainment of the Hájek-Le Cam bound for truncated squared geodesic loss are laid out in global normal coordinates, and they include a spectral-clipping estimator plus Monte Carlo checks that line up with the asymptotics. That package is useful for anyone needing rigorous profile inference on this manifold model. The soft spot is the uniform integrability condition required to extend the minimax attainment to ordinary squared geodesic loss. Hyperbolic geometry has exponential volume growth, so squared distances can have heavier tails under local perturbations even with eigenvalues bounded away from zero and infinity; the paper asserts the condition holds but does not appear to supply a separate tail bound or moment restriction to confirm uniformity. The interior-shell assumption is standard but narrows applicability. This is for readers working on statistical inference for manifold-valued or hyperbolic data, such as in network science or hierarchical modeling. It deserves peer review because the core derivations follow standard likelihood and information-matrix steps and the gaps are technical rather than structural.

Referee Report

2 major / 2 minor

Summary. The paper develops profile likelihood inference for the anisotropic hyperbolic wrapped normal distribution on hyperbolic space. It formulates the estimator on a covariance shell to ensure finite-sample existence, proves well-posedness of the constrained likelihood, identifiability of the family, and strong consistency when the true covariance is interior to the shell. In global normal coordinates for location and log-covariance coordinates for the nuisance parameter, it establishes joint asymptotic normality, derives the efficient profile information matrix via the Schur complement, proves local asymptotic normality of the experiment, and obtains the Hájek–Le Cam local asymptotic minimax lower bound under squared geodesic loss. The profile estimator is shown to attain this bound for truncated squared loss and for ordinary squared loss under a uniform-integrability condition. An explicit computational form based on spectral clipping of the empirical tangent covariance is given, together with a Monte Carlo study of finite-sample risk and coverage.

Significance. If the technical conditions hold, the work supplies a complete asymptotic theory for efficient inference on the location parameter of a flexible non-Euclidean model that arises in directional statistics and geometric data analysis. The combination of manifold geometry, profile likelihood, LAN, and an explicit minimax bound is a solid contribution; the computational recipe and simulation calibration further support practical use.

major comments (2)

[section on local asymptotic normality and minimax bounds] The attainment of the Hájek–Le Cam lower bound under ordinary (untruncated) squared geodesic loss, stated after the LAN result, rests on an unverified uniform-integrability condition. Because of the exponential volume growth induced by negative curvature, the tails of d_g²(μ̂_n, μ) under local perturbations of order n^{-1/2} may fail to be uniformly integrable even when covariance eigenvalues are bounded away from zero and infinity; no separate moment bound or truncation argument is supplied to confirm that E[d_g²(μ̂_n, μ) 1_{d_g > M}] → 0 uniformly in the local parameter.
[consistency theorem] The strong-consistency statement requires the true covariance to lie in the interior of the covariance shell. The paper does not discuss the behavior of the estimator when the true parameter approaches the boundary of the shell or when the shell radius is chosen data-dependently; this leaves open whether consistency can be extended or whether the finite-sample constraint introduces asymptotic bias in near-boundary regimes.

minor comments (2)

[model definition] Notation for the hyperbolic distance d_g and the tangent-space norm should be introduced once with a clear reference to the model definition rather than repeated inline.
[simulation section] The Monte Carlo study reports scaled location risk and Wald coverage; adding a brief statement on the number of replications and the range of covariance eigenvalues used would improve reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading, positive assessment of the contributions, and constructive major comments. We address each point below and will make targeted revisions to clarify limitations and strengthen the presentation.

read point-by-point responses

Referee: [section on local asymptotic normality and minimax bounds] The attainment of the Hájek–Le Cam lower bound under ordinary (untruncated) squared geodesic loss, stated after the LAN result, rests on an unverified uniform-integrability condition. Because of the exponential volume growth induced by negative curvature, the tails of d_g²(μ̂_n, μ) under local perturbations of order n^{-1/2} may fail to be uniformly integrable even when covariance eigenvalues are bounded away from zero and infinity; no separate moment bound or truncation argument is supplied to confirm that E[d_g²(μ̂_n, μ) 1_{d_g > M}] → 0 uniformly in the local parameter.

Authors: We acknowledge that the manuscript relies on an unverified uniform-integrability condition to extend the minimax attainment from truncated to ordinary squared geodesic loss. The current proof establishes the result for the truncated loss without additional assumptions and invokes the condition for the untruncated case, without supplying explicit moment bounds or truncation arguments to confirm uniform integrability. Given the exponential volume growth in hyperbolic space, such verification is technically involved and is not carried out here. In the revised version we will rephrase the statement to present the truncated-loss result as the unconditional achievement and the ordinary-loss result as conditional on the stated integrability assumption, while adding a remark that explicit verification of the condition under bounded-eigenvalue assumptions remains an open technical question. revision: partial
Referee: [consistency theorem] The strong-consistency statement requires the true covariance to lie in the interior of the covariance shell. The paper does not discuss the behavior of the estimator when the true parameter approaches the boundary of the shell or when the shell radius is chosen data-dependently; this leaves open whether consistency can be extended or whether the finite-sample constraint introduces asymptotic bias in near-boundary regimes.

Authors: We agree that strong consistency is proved only when the true covariance lies in the interior of the shell, which ensures the constraint is asymptotically non-binding. The manuscript does not analyze the boundary regime or data-dependent shell radii. In the revision we will add a short paragraph after the consistency theorem noting that, as the true covariance approaches the boundary, the estimator may converge to a boundary point of the shell and that data-dependent choices of the radius would require a separate argument to preclude asymptotic bias. The main theoretical development focuses on the interior regime where the guarantees are cleanest. revision: yes

Circularity Check

0 steps flagged

No significant circularity

full rationale

The derivation chain proceeds from the definition of the anisotropic hyperbolic wrapped normal density, through standard manifold coordinate charts (global normal coordinates for location, log-covariance for nuisance), to the profile likelihood, LAN property via standard Le Cam theory, and the Hájek-Le Cam lower bound under squared geodesic loss. These steps invoke general results on manifold MLEs and information matrices rather than any self-referential definition, fitted parameter renamed as prediction, or load-bearing self-citation. The uniform-integrability condition is explicitly stated as an extra assumption for the non-truncated loss case and is not derived by reduction to the model inputs. The paper is self-contained against external benchmarks from differential geometry and asymptotic statistics.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claims rest on standard properties of Riemannian manifolds, the hyperbolic space model, and classical likelihood theory; no free parameters are introduced beyond the model definition, and no new entities are postulated.

axioms (2)

standard math Hyperbolic space is a complete Riemannian manifold with constant negative curvature, admitting global normal coordinates and tangent space identification.
Invoked throughout for the definition of the wrapped normal distribution and geodesic loss.
domain assumption The wrapped normal family is identifiable and the likelihood is sufficiently regular for profile estimation and asymptotic expansions.
Required for the consistency and local asymptotic normality results stated in the abstract.

pith-pipeline@v0.9.0 · 5527 in / 1503 out tokens · 51823 ms · 2026-05-09T19:01:37.492799+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

300 extracted references · 216 canonical work pages · 4 internal anchors

[1]

Positive

Bhatia, Rajendra , month = dec, year =. Positive. doi:10.1515/9781400827787 , urldate =

work page doi:10.1515/9781400827787
[2]

Medians and

Arnaudon, Marc and Barbaresco, Frédéric and Yang, Le , editor =. Medians and. Matrix. 2013 , note =. doi:10.1007/978-3-642-30232-9_8 , language =

work page doi:10.1007/978-3-642-30232-9_8 2013
[3]

Econometrics and Statistics , author =

Modeling. Econometrics and Statistics , author =. 2022 , pages =. doi:10.1016/j.ecosta.2021.04.004 , language =

work page doi:10.1016/j.ecosta.2021.04.004 2022
[4]

Riemannian

Petersen, Peter , year =. Riemannian. doi:10.1007/978-0-387-29403-2 , publisher =

work page doi:10.1007/978-0-387-29403-2
[5]

You, Kisung , year =. Finite. doi:10.48550/ARXIV.2604.24895 , abstract =

work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2604.24895
[6]

The Annals of Statistics , author =

Fréchet regression for random objects with. The Annals of Statistics , author =. doi:10.1214/17-AOS1624 , number =

work page doi:10.1214/17-aos1624
[7]

NeuroImage , author =

The. NeuroImage , author =. 2012 , pages =. doi:10.1016/j.neuroimage.2012.02.018 , language =

work page doi:10.1016/j.neuroimage.2012.02.018 2012
[8]

and Pedersen, M.S

Petersen, K.B. and Pedersen, M.S. , month = nov, year =. The
[9]

Science , author =

Prediction of. Science , author =. 2010 , pages =. doi:10.1126/science.1194144 , language =

work page doi:10.1126/science.1194144 2010
[10]

A wrapped normal distribution on hyperbolic space for gradient-based learning , volume =

Nagano, Yoshihiro and Yamaguchi, Shoichiro and Fujita, Yasuhiro and Koyama, Masanori , editor =. A wrapped normal distribution on hyperbolic space for gradient-based learning , volume =. Proceedings of the 36th international conference on machine learning , publisher =. 2019 , pages =

2019
[11]

, year =

Lee, John M. , year =. Riemannian
[12]

Springer Series in Statistics

Le Cam, Lucien and Lo Yang, Grace , year =. Asymptotics in. doi:10.1007/978-1-4612-1166-2 , publisher =

work page doi:10.1007/978-1-4612-1166-2
[13]

A General Qualitative Definition of Robustness

Asymptotic. The Annals of Mathematical Statistics , author =. 1971 , pages =. doi:10.1214/aoms/1177693066 , language =

work page doi:10.1214/aoms/1177693066 1971
[14]

The Annals of Mathe- matical Statistics22(1), 79–86 (1951) https://doi.org/10.1214/aoms/1177729694

Note on the. The Annals of Mathematical Statistics , author =. 1949 , pages =. doi:10.1214/aoms/1177729952 , language =

work page doi:10.1214/aoms/1177729952 1949
[15]

Local asymptotic minimax and admissibility in estimation , language =

Hajek, Jaroslav , year =. Local asymptotic minimax and admissibility in estimation , language =. Proceedings of the
[16]

, year =

Cox, D.R. , year =. Analysis of. doi:10.1201/9781315137391 , urldate =

work page doi:10.1201/9781315137391
[17]

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

An. Journal of the Royal Statistical Society Series B: Statistical Methodology , author =. 1964 , pages =. doi:10.1111/j.2517-6161.1964.tb00553.x , abstract =

work page doi:10.1111/j.2517-6161.1964.tb00553.x 1964
[18]

Computational Statistics & Data Analysis , author =

Parameter estimation and model-based clustering with spherical normal distribution on the unit hypersphere , volume =. Computational Statistics & Data Analysis , author =. 2022 , pages =. doi:10.1016/j.csda.2022.107457 , urldate =

work page doi:10.1016/j.csda.2022.107457 2022
[19]

, month = nov, year =

Li, Songzi and Rizzo, Maria L. , month = nov, year =. K-groups:. doi:10.48550/arXiv.1711.04359 , abstract =

work page doi:10.48550/arxiv.1711.04359
[20]

Directional

Hauberg, Soren , month = jul, year =. Directional. 2018 21st. doi:10.23919/ICIF.2018.8455242 , urldate =

work page doi:10.23919/icif.2018.8455242 2018
[21]

Hyper- spherical variational auto-encoders.arXiv preprint arXiv:1804.00891, 2018

Davidson, Tim R. and Falorsi, Luca and Cao, Nicola De and Kipf, Thomas and Tomczak, Jakub M. , month = sep, year =. Hyperspherical. doi:10.48550/arXiv.1804.00891 , abstract =

work page doi:10.48550/arxiv.1804.00891
[22]

2022 , note =

Directional. 2022 , note =. doi:10.1007/978-981-19-1044-9 , language =

work page doi:10.1007/978-981-19-1044-9 2022
[23]

On the spherical

You, Kisung and Shung, Dennis , year =. On the spherical. doi:10.48550/ARXIV.2208.11929 , abstract =

work page doi:10.48550/arxiv.2208.11929
[24]

Physical Review E , author =

Fuzzy communities and the concept of bridgeness in complex networks , volume =. Physical Review E , author =. 2008 , pages =. doi:10.1103/PhysRevE.77.016107 , number =

work page doi:10.1103/physreve.77.016107 2008
[25]

Journal of Anthropological Research , author =

An. Journal of Anthropological Research , author =. 1977 , pages =. doi:10.1086/jar.33.4.3629752 , number =

work page doi:10.1086/jar.33.4.3629752 1977
[26]

Journal of Complex Networks , author =

Hydra: a method for strain-minimizing hyperbolic embedding of network- and distance-based data , volume =. Journal of Complex Networks , author =. 2020 , pages =. doi:10.1093/comnet/cnaa002 , number =

work page doi:10.1093/comnet/cnaa002 2020
[27]

Numerische Mathematik , author =

A note on two problems in connexion with graphs , volume =. Numerische Mathematik , author =. 1959 , pages =. doi:10.1007/BF01386390 , number =

work page doi:10.1007/bf01386390 1959
[28]

Foundations of Data Science , author =

Tverberg's theorem and multi-class support vector machines , volume =. Foundations of Data Science , author =. 2025 , pages =. doi:10.3934/fods.2024039 , number =

work page doi:10.3934/fods.2024039 2025
[29]

Biometrika , author =

Multivariate output analysis for. Biometrika , author =. 2019 , pages =. doi:10.1093/biomet/asz002 , language =

work page doi:10.1093/biomet/asz002 2019
[30]

Biometrika , author =

Fréchet analysis of variance for random objects , volume =. Biometrika , author =. 2019 , pages =. doi:10.1093/biomet/asz052 , abstract =

work page doi:10.1093/biomet/asz052 2019
[31]

Proceedings of the American Mathematical Society , author =

Omnibus. Proceedings of the American Mathematical Society , author =. 2016 , pages =. doi:10.1090/proc/13216 , number =

work page doi:10.1090/proc/13216 2016
[32]

, month = may, year =

Ginestet, Cedric E. , month = may, year =. Strong
[33]

Journal of Multivariate Analysis , author =

The. Journal of Multivariate Analysis , author =. 1982 , pages =. doi:10.1016/0047-259X(82)90077-X , language =

work page doi:10.1016/0047-259x(82)90077-x 1982
[34]

Annales de l'institut Henri Poincaré , author =

Les éléments aléatoires de nature quelconque dans un espace distancié , volume =. Annales de l'institut Henri Poincaré , author =. 1948 , pages =

1948
[35]

Boumal, Nicolas , year =. An. doi:10.1017/9781009166164 , language =

work page doi:10.1017/9781009166164
[36]

Mishne, Gal and Wan, Zhengchao and Wang, Yusu and Yang, Sheng , year =. The. Proceedings of the 40th
[37]

and Bedo, Justin , month = jun, year =

Song, Le and Smola, Alex and Gretton, Arthur and Borgwardt, Karsten M. and Bedo, Justin , month = jun, year =. Supervised feature selection via dependence estimation , isbn =. Proceedings of the 24th international conference on. doi:10.1145/1273496.1273600 , language =

work page doi:10.1145/1273496.1273600
[38]

Dryden, I. L. and Mardia, K. V. , year =. Statistical shape analysis , isbn =
[39]

Dryden, I. L. and Mardia, K. V. , year =. Statistical shape analysis with applications in
[40]

Mardia, K. V. and Jupp, Peter E. , year =. Directional statistics , isbn =
[41]

Maximum Likelihood from Incomplete Data Via the EM Algorithm,

Maximum. Journal of the Royal Statistical Society: Series B (Methodological) , author =. 1977 , pages =. doi:10.1111/j.2517-6161.1977.tb01600.x , number =

work page doi:10.1111/j.2517-6161.1977.tb01600.x 1977
[42]

Differentiating through the

Lou, Aaron and Katsman, Isay and Jiang, Qingxuan and Belongie, Serge and Lim, Ser-Nam and De Sa, Christopher , year =. Differentiating through the. Proceedings of the 37th
[43]

Qiu, Rui and Yu, Zhou and Zhu, Ruoqing , month = jan, year =. Random. Journal of Machine Learning Research , publisher =
[44]

Computational Statistics & Data Analysis , author =

Medoid splits for efficient random forests in metric spaces , volume =. Computational Statistics & Data Analysis , author =. 2024 , pages =. doi:10.1016/j.csda.2024.107995 , language =

work page doi:10.1016/j.csda.2024.107995 2024
[45]

Journal of Statistical Planning and Inference , author =

Design and relative efficiency in two-phase studies , volume =. Journal of Statistical Planning and Inference , author =. 2012 , pages =. doi:10.1016/j.jspi.2012.04.013 , language =

work page doi:10.1016/j.jspi.2012.04.013 2012
[46]

Riemannian radial distributions on

Chen, Hengchao , year =. Riemannian radial distributions on. doi:10.48550/ARXIV.2405.07852 , abstract =

work page doi:10.48550/arxiv.2405.07852
[47]

IEEE Transactions on Information Theory , author =

Gaussian. IEEE Transactions on Information Theory , author =. 2018 , pages =. doi:10.1109/TIT.2017.2713829 , number =

work page doi:10.1109/tit.2017.2713829 2018
[48]

Using early rejection markov chain monte carlo and gaussian processes to accelerate abc methods

Bayesian. Journal of Computational and Graphical Statistics , author =. 2024 , pages =. doi:10.1080/10618600.2024.2308219 , language =

work page doi:10.1080/10618600.2024.2308219 2024
[49]

Scientific Reports , author =

Navigability of. Scientific Reports , author =. 2017 , pages =. doi:10.1038/s41598-017-08872-4 , abstract =

work page doi:10.1038/s41598-017-08872-4 2017
[50]

Sarkar, Rik , editor =. Low. 2012 , pages =. doi:10.1007/978-3-642-25878-7_34 , booktitle =

work page doi:10.1007/978-3-642-25878-7_34 2012
[51]

Learning continuous hierarchies in the

Nickel, Maximillian and Kiela, Douwe , editor =. Learning continuous hierarchies in the. Proceedings of the 35th. 2018 , pages =

2018
[52]

Poincaré embeddings for learning hierarchical representations , volume =

Nickel, Maximillian and Kiela, Douwe , editor =. Poincaré embeddings for learning hierarchical representations , volume =. Advances in
[53]

Journal of the American Statistical Association , author =

Hyperbolic. Journal of the American Statistical Association , author =. 2026 , pages =. doi:10.1080/01621459.2026.2635077 , language =

work page doi:10.1080/01621459.2026.2635077 2026
[54]

, collaborator =

Abramowitz, Milton and Stegun, Irene A. , collaborator =. Handbook of
[55]

Acemoglu, Daron and Autor, David , year =. Skills,. Handbook of. doi:10.1016/S0169-7218(11)02410-5 , language =

work page doi:10.1016/s0169-7218(11)02410-5
[56]

American Economic Review , author =

The. American Economic Review , author =. 2018 , pages =. doi:10.1257/aer.20160696 , abstract =

work page doi:10.1257/aer.20160696 2018
[57]

An in depth look at the

Adamo, Davide and Corneli, Marco and Vuillien, Manon and Vila, Emmanuelle , year =. An in depth look at the. Forty-second
[58]

doi:10.18637/jss.v050.i05 , language =

Journal of Statistical Software , author =. doi:10.18637/jss.v050.i05 , language =

work page doi:10.18637/jss.v050.i05
[59]

Proceedings of the American Mathematical Society , author =

Riemannian \. Proceedings of the American Mathematical Society , author =. 2011 , pages =. doi:10.1090/S0002-9939-2010-10541-5 , language =

work page doi:10.1090/s0002-9939-2010-10541-5 2011
[60]

SIAM Journal on Control and Optimization , author =

On the. SIAM Journal on Control and Optimization , author =. 2013 , pages =. doi:10.1137/12086282X , language =

work page doi:10.1137/12086282x 2013
[61]

IEEE Transactions on Pattern Analysis and Machine Intelligence , author =

Generalized. IEEE Transactions on Pattern Analysis and Machine Intelligence , author =. 2015 , pages =. doi:10.1109/TPAMI.2014.2353625 , number =

work page doi:10.1109/tpami.2014.2353625 2015
[62]

and Chowdhury, A.K.R

Aggarwal, G. and Chowdhury, A.K.R. and Chellappa, R. , year =. A system identification approach for video-based face recognition , isbn =. Proceedings of the 17th. doi:10.1109/ICPR.2004.1333732 , urldate =

work page doi:10.1109/icpr.2004.1333732 2004
[63]

Barycenters in the

Barycenters in the. SIAM Journal on Mathematical Analysis , author =. 2011 , pages =. doi:10.1137/100805741 , language =

work page doi:10.1137/100805741 2011
[64]

Information

Akaike, Hirotogu , editor =. Information. 1998 , pages =. doi:10.1007/978-1-4612-1694-0_15 , booktitle =

work page doi:10.1007/978-1-4612-1694-0_15 1998
[65]

Data Mining and Knowledge Discovery , author =

Extreme-value-theoretic estimation of local intrinsic dimensionality , volume =. Data Mining and Knowledge Discovery , author =. 2018 , pages =. doi:10.1007/s10618-018-0578-6 , number =

work page doi:10.1007/s10618-018-0578-6 2018
[66]

Zico , editor =

Amos, Brandon and Xu, Lei and Kolter, J. Zico , editor =. Input convex neural networks , volume =. Proceedings of the 34th international conference on machine learning , publisher =. 2017 , pages =

2017
[67]

Scientific Reports , author =

The quest for identifiability in human functional connectomes , volume =. Scientific Reports , author =. 2018 , pages =. doi:10.1038/s41598-018-25089-1 , language =

work page doi:10.1038/s41598-018-25089-1 2018
[68]

Gradient flows: in metric spaces and in the space of probability measures , isbn =

Ambrosio, Luigi and Gigli, Nicola and Savaré, Giuseppe , year =. Gradient flows: in metric spaces and in the space of probability measures , isbn =
[69]

Lectures on

Ambrosio, Luigi and Brué, Elia and Semola, Daniele , year =. Lectures on. doi:10.1007/978-3-72162-6 , language =

work page doi:10.1007/978-3-72162-6
[70]

2003 , keywords =

Optimal transportation and applications: lectures given at the. 2003 , keywords =

2003
[71]

Journal of Optimization Theory and Applications , author =

Weiszfeld’s. Journal of Optimization Theory and Applications , author =. 2015 , pages =. doi:10.1007/s10957-014-0586-7 , language =

work page doi:10.1007/s10957-014-0586-7 2015
[72]

Laplacian eigenmaps and spectral techniques for embedding and clustering , volume =

Belkin, Mikhail and Niyogi, Partha , editor =. Laplacian eigenmaps and spectral techniques for embedding and clustering , volume =. Advances in neural information processing systems , publisher =
[73]

Convergence of laplacian eigenmaps , volume =

Belkin, Mikhail and Niyogi, Partha , editor =. Convergence of laplacian eigenmaps , volume =. Advances in neural information processing systems , publisher =
[74]

The Annals of Applied Statistics , author =

A novel estimator of. The Annals of Applied Statistics , author =. doi:10.1214/23-AOAS1802 , number =

work page doi:10.1214/23-aoas1802
[75]

and Krim, H

Ben Hamza, A. and Krim, H. , year =. Jensen-renyi divergence measure: theoretical and computational perspectives , isbn =. doi:10.1109/ISIT.2003.1228271 , urldate =

work page doi:10.1109/isit.2003.1228271 2003
[76]

Numerische Mathematik , author =

A computational fluid mechanics solution to the. Numerische Mathematik , author =. 2000 , pages =. doi:10.1007/s002110050002 , number =

work page doi:10.1007/s002110050002 2000
[77]

SIAM Journal on Scientific Computing , author =

Iterative. SIAM Journal on Scientific Computing , author =. 2015 , pages =. doi:10.1137/141000439 , language =

work page doi:10.1137/141000439 2015
[78]

, month = nov, year =

Bendokat, Thomas and Zimmermann, Ralf and Absil, P.-A. , month = nov, year =. A
[79]

Journal of the Royal Statistical Society: Series B (Methodological) , volume = 57, number = 1, pages =

Controlling the. Journal of the Royal Statistical Society Series B: Statistical Methodology , author =. 1995 , pages =. doi:10.1111/j.2517-6161.1995.tb02031.x , language =

work page doi:10.1111/j.2517-6161.1995.tb02031.x 1995
[80]

1984 , PAGES =

Berg, Christian and Christensen, Jens Peter Reus and Ressel, Paul , year =. Harmonic. doi:10.1007/978-1-4612-1128-0 , urldate =

work page doi:10.1007/978-1-4612-1128-0

Showing first 80 references.