Pith. sign in

REVIEW 4 major objections 5 minor 51 references

Latent Manifold Reconstruction and Representation with Topological and Geometrical Regularization

T0 review · 4 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read An autoencoder with a manifold reconstruction layer and topological and geometric regularizers produces low-dimensional embeddings of noisy point clouds that preserve both global shape and local detail, outperforming t-SNE, UMAP, and…

desk verdict A credible integration of existing manifold-learning pieces with attractive visuals, but the abstract's outperformance claim is contradicted by the paper's own Table 1 and the trainable-radius mechanism is under-specified. read the letter →

arxiv 2505.04412 v1 pith:6FACFEAB submitted 2025-05-07 cs.LG

Ren Wang , Pengcheng Zhou This is my paper · ORCID
classification cs.LG
keywords manifoldlearningdimensionalityreductionautoencoderpersistenthomologytopologicalregularizationgeometricpointcloudsreconstruction
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

This paper claims that adding a learned manifold reconstruction step to an autoencoder lets dimensionality reduction preserve both the global shape and the local geometry of noisy point clouds. The central idea is to first contract noisy points toward the latent manifold with a reconstruction layer, then train the encoder with two regularizers: one that matches the persistent-homology signatures of the reconstructed manifold and the embedding, and one that penalizes distortion measured against a scaled isometry. The authors report that this combined system beats standard baselines on noisy Swiss roll, mammoth, 100-dimensional spheres, and object part point clouds, and that the reconstruction and representation components improve each other during training. If the claim is right, a single end-to-end pipeline can denoise, reconstruct, and meaningfully embed noisy Euclidean data without losing the shape that makes the data interpretable.

What carries the argument

The central object is the Manifold Reconstruction Layer (MRL), a map $F_M$ from the ambient Euclidean space to itself that estimates the latent manifold by local contraction: for each point $x$, it computes a weighted mean direction toward nearby points and repositions $x$ inside a cylinder aligned with that direction, using the smooth weight functions of Eqs. (4) and (5). Its output $Y$ feeds both the autoencoder and the two regularizers. The topological regularizer uses persistent homology, comparing distance entries of $Y$ and $Z$ selected by the persistent pairings of the Vietoris-Rips filtration, via a topological signature loss. The geometric regularizer evaluates the eigenvalues of the pulled-back metric $J^T H J$ relative to the metric on $Y$ and penalizes deviations from a scaled isometry. Together the two regularizers carry the paper's claim that global topology and local geometry can be preserved simultaneously.

What would settle it

Freeze the MRL radii $r_0, r_1, r_2$ at their initialization values, retrain on the Swiss roll dataset, and compare reconstruction error and embedding metrics with the version that back-propagates through the radii; if the two are indistinguishable, the claim that trained radii drive the mutual-promotion effect is falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that a low-dimensional embedding of a noisy Euclidean point cloud can simultaneously preserve the global topology and the local geometry of the latent manifold, provided the raw points are first contracted toward that manifold by a trainable Manifold Reconstruction Layer. The layer estimates, for each point, the direction toward the closest manifold point by a weighted average of neighbors in a ball of radius $r_0$, then moves the point along that direction using a cylinder-shaped neighborhood with radii $r_1$ and $r_2$; the three radii are claimed to be optimized through back-propagation. The reconstructed cloud $Y$ is then encoded to $Z$, and two regularizers are applied between $Y$ and $Z$: a topological signature loss comparing persistent diagrams of $Y$ and $Z$, and a geometric loss based on a relaxed distortion measure that encourages the encoder to be a scaled isometry. The total loss is a weighted sum of the autoencoder reconstruction loss, $L_{topo}$, and $L_{geom}$. The paper argues that this configuration makes the reconstruction and representation tasks mutually beneficial, and presents experiments on four datasets where its embeddings retain shapes that t-SNE, UMAP, and Topological AutoEncoders fracture or flatten.

Load-bearing premise

The load-bearing premise is that the three radius parameters controlling how far the reconstruction layer looks are actually trainable: the weight functions in the layer contain hard cutoffs, and the paper does not say how gradients pass through those cutoffs, so if the radii stay at their initial values the claimed mutual promotion of reconstruction and representation lacks its mechanism.

Editorial extensions

If this is right

  • If the claim is correct, noisy point clouds can be embedded without a separate denoising pass; the reconstruction layer performs denoising and manifold estimation in the same end-to-end training.
  • The results imply that a topological regularizer alone is not enough to prevent folded, self-overlapping embeddings; the geometric term is needed to preserve local proportions, and the two regularizers are complementary.
  • The ablation evidence supports the mutual-promotion claim: the reconstructed manifold is cleaner when the autoencoder and regularizers are present, and the embedding improves when the reconstruction layer is present.
  • The same architecture can in principle be applied to any Euclidean point cloud input, and the supplementary robustness experiments suggest it tolerates a range of noise levels, sample sizes, and input dimensions without changing the training setup.

Reading between the lines

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

  • If the radii do not actually receive gradients because of the hard cutoffs in Eqs. (4) and (5), the reported gain may come from the reconstruction layer acting as a fixed denoiser; freezing the radii in an ablation would separate the two explanations.
  • The topological regularizer's cost grows with the Vietoris-Rips complex size, so for very large point clouds a practical extension is to compute persistent homology on landmark subsamples and compare signatures on those landmarks.
  • Because the geometric regularizer enforces approximate scaled isometry, the method may need curvature-aware extensions on strongly curved or high-intrinsic-dimension manifolds; the paper does not test this regime.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. This paper proposes an autoencoder-based manifold learning method that prepends a Manifold Reconstruction Layer (MRL) to the encoder and adds topological and geometric regularizers computed between the reconstructed manifold and the low-dimensional embedding. The MRL contracts noisy points toward an estimated latent manifold following Yao et al. [8]; the topological regularizer is the topological signature loss of Moor et al. [6]; the geometric regularizer is a relaxed distortion measure. The total loss is a weighted combination of an autoencoder reconstruction term, the topological loss, and the geometric loss. The authors claim that the manifold reconstruction and representation components promote each other during end-to-end training, and that experiments on Swiss roll, mammoth, PartNet, and spheres datasets show the method outperforms baselines such as t-SNE, UMAP, and Topological AutoEncoders. The supplement contains ablations, noise robustness experiments, scaling experiments, and a hyperparameter grid search.

Significance. If the central claims were fully supported, the contribution would be useful: an end-to-end model that balances global topology and local geometry in dimensionality reduction while also denoising the input through manifold reconstruction. The paper ships a public code repository, reports metrics averaged over 10 embeddings, and includes ablation and robustness studies, which are strengths. However, the quantitative evidence in Table 1 does not support the abstract's outperformance claim, the trainability of the MRL radii is not established, and the main formula for the geometric regularizer is asserted without derivation. The idea of coupling manifold fitting with representation learning is promising, but the validation and the theoretical presentation need substantial work before the central claims can be accepted.

major comments (4)
  1. [Section 4.5, Table 1] Table 1 contradicts the abstract's claim that the method 'outperforms baselines like t-SNE, UMAP, and Topological AutoEncoders.' Taking the standard direction of each metric (lower is better for KL and RMSE; higher is better for kNN, Trust, and Spearman), the proposed method wins only 9 of 24 metric-dataset cells; t-SNE wins 10 cells and Topological AutoEncoder wins 6 cells, and on the Spheres dataset the proposed method wins none of the six metrics. No aggregation rule, weighting scheme, or significance test is specified that would convert this partial record into an overall win. This is load-bearing: the abstract and Section 4.4 present quantitative validation as support for the central claim, but the paper's own table does not support that claim.
  2. [Section 3.3, Eqs. (4)-(5)] The statement that r0, r1, and r2 are 'passed through back-propagation and optimized in the training process' is not supported by the definitions. The weights in Eqs. (4) and (5) contain hard cutoffs such as '0, otherwise,' and the neighbor sets I_x and J_x depend on the radii through ball-membership conditions, so the mapping from the radii to the MRL output is not differentiable with respect to the radii in the standard sense. The paper does not describe any relaxation, smoothing, or straight-through estimator. If the radii are in fact fixed, the claimed mutual promotion between the MRL and the autoencoder loses its stated mechanism; if they are optimized, the optimization must be specified.
  3. [Section 3.5, Eq. (10)] The closed form L_geom = D/2 * E[Tr(H^2)] / E[Tr(H)]^2 - D is asserted without derivation. Section 3 explicitly states that most mathematical derivations are omitted, and Definition 2 says the formula follows from choosing a specific convex function h and symmetric function S in Eq. (9), but no such choice is given. Since the geometric regularizer is one of the two core regularizers, the equality needs to be derived or a precise pointer to the reference (e.g., [26] or [10]) with the exact h and S must be stated. As written, the formula cannot be verified from the manuscript.
  4. [Section 5, Table S2] The 'mutual promotion' conclusion is not cleanly identified by the ablation. The comparison of 'Final model' versus 'MR AE' in the 'Point Cloud vs Manifold' group shows improvement, but the Final model differs from MR AE by both the topological regularizer and the geometric regularizer in addition to the autoencoder's reconstruction feedback, so the improvement cannot be attributed specifically to 'the representation component (the AutoEncoder).' The additional comparison between 'Manifold vs Embedding' for the Final model and 'Point Cloud vs Embedding' for Topo-geom AE compares different quantities and lacks a controlled baseline; it is suggestive but not conclusive. The paper should either add an ablation that isolates the autoencoder's feedback to the MRL or soften the mutual-promotion claim.
minor comments (5)
  1. [Section 5] The unrelated aphorism about 'clearer answers' and 'clearer writing' interrupts the ablation discussion and should be removed; it is not a scientific statement and is not previously claimed in the paper.
  2. [Appendix C.1, Definition C1] Definition C1 states 'Positivity: |v| ≤ 0,' which should be '|v| ≥ 0.'
  3. [Appendix C.1, Definition C2] The support of a function is written with an equality sign on the right-hand side; the closure should be indicated with an overline or explicitly stated, since the text says it is the closure.
  4. [Table 1] Table 1 contains apparent notation typos, such as '503 × 10^{-1}' for t-SNE RMSE on Swiss roll and '114 × 10^{-1}' for UMAP on the same row, which would be implausible values; these entries should be corrected and the table formatting checked.
  5. [Appendix E.3.3] The paragraph on scaling to higher-dimensional data says 'the input sample sizes of the Swiss roll dataset' in the context of the Spheres dataset; it should refer to the input dimensions of the Spheres dataset.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the components are independently cited, and the claims are tested against external baselines.

full rationale

I find no significant circularity in this paper. The method is an integration of independently published components: the Manifold Reconstruction Layer is based on Yao et al. [8], the topological regularizer on Moor et al. [6], and the geometric regularizer on Lee et al. [26]. These are external citations, not self-citations by the authors. The only imported theoretical result, Theorem C1, is explicitly attributed to the manifold fitting algorithm of Yao et al. and is not derived from the paper's own claims. The paper's contribution is the combined loss in Eq. (11) and its empirical evaluation, not a first-principles derivation of a result from an equivalent input. The evaluation is conducted against external baselines (t-SNE, UMAP, Topological AutoEncoder, PCA, Isomap) on external datasets, so the central claim of improved manifold discovery and preservation is empirically falsifiable and not enforced by construction. The MRL radii are hyperparameters that are optimized, but they are not fitted to a subset and then reported as predictions of the same quantity. Even the concern about non-differentiability of the radii is an implementation-correctness issue, not circularity. The abstract's 'outperforms' claim may be challenged by the win counts in Table 1, but that is a quantitative-evidence or correctness concern, not a circularity concern. There are no self-citations by the authors, and the unusual phrase 'as we previously claimed' is not load-bearing for any derivation. The paper is self-contained against external benchmarks, so the honest finding is no circularity.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

No new entities are postulated. The free parameters are the MRL radii and the per-dataset loss weights; the radii are claimed trainable but no gradients or learned values are provided, and the loss weights are tuned per dataset. The important unstated assumption is that the hard-cutoff weighting functions remain optimizable under back-propagation.

free parameters (5)
  • r0 (initial value) = 1.0
    MRL ball radius for contraction direction; initialized to 1.0 and claimed to be optimized via back-propagation, but no learned values or gradient treatment are reported (Section 3.3, Appendix D.2).
  • r1 (initial value) = 0.01
    Cylinder radius perpendicular to contraction direction; initialized per experiment, claimed trainable, final values unspecified.
  • r2 (initial value) = 1.0
    Cylinder radius along contraction direction; initialized to 1.0, claimed trainable, final values unspecified.
  • smoothness k = 3
    Fixed integer chosen for twice-differentiability of the weighting functions (Eq. 4-5); no sensitivity analysis.
  • loss weights (lambda_AE, lambda_topo, lambda_geom) = per-dataset, e.g., (1,1,5) for Swiss roll
    Chosen separately for each dataset in Section 4.3; the grid search in Appendix E.4 shows metrics are sensitive to these weights, so the reported results depend on hand tuning.
assumptions (4)
  • domain assumption Assumptions 1-4: ambient space is Euclidean, noise is Gaussian with zero mean, latent manifold is twice-differentiable and compact, data distribution is uniform on the manifold.
    Stated in Section 3.2. The manifold reconstruction layer's contraction estimates rely on these; real datasets like PartNet may violate the uniform and Gaussian assumptions.
  • domain assumption Theorem C1: the manifold fitting map FM approximates the projection onto the manifold with error O(sigma^2 log(1/sigma)) with high probability.
    This bound is imported from Yao et al. (Ref [8]) and restated as Theorem C1 without proof; the method's denoising behavior depends on it.
  • ad hoc to paper The relaxed distortion measure formula (Eq. 10) follows from choosing a specific convex h and symmetric S in Eq. 9.
    The specific functions that turn Eq. 9 into Eq. 10 are not stated; the formula is taken from prior work [10, 26] but the paper presents it as its own derivation without details.
  • domain assumption Persistence pairings of the reconstructed manifold Y are informative for the true manifold.
    The topological regularizer is computed between Y and Z; if Y is a poor estimate of the true manifold, the regularizer enforces the wrong topology (Sections 3.4, 5).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Latent Manifold Reconstruction and Representation with Topological and Geometrical Regularization." pith.science (2026). https://pith.science/paper/6FACFEAB

@misc{pith2026250504412,
  author       = {Pith},
  title        = {Pith review of: Latent Manifold Reconstruction and Representation with Topological and Geometrical Regularization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6FACFEAB}},
  note         = {Machine review of arXiv:2505.04412}
}
read the original abstract

Manifold learning aims to discover and represent low-dimensional structures underlying high-dimensional data while preserving critical topological and geometric properties. Existing methods often fail to capture local details with global topological integrity from noisy data or construct a balanced dimensionality reduction, resulting in distorted or fractured embeddings. We present an AutoEncoder-based method that integrates a manifold reconstruction layer, which uncovers latent manifold structures from noisy point clouds, and further provides regularizations on topological and geometric properties during dimensionality reduction, whereas the two components promote each other during training. Experiments on point cloud datasets demonstrate that our method outperforms baselines like t-SNE, UMAP, and Topological AutoEncoders in discovering manifold structures from noisy data and preserving them through dimensionality reduction, as validated by visualization and quantitative metrics. This work demonstrates the significance of combining manifold reconstruction with manifold learning to achieve reliable representation of the latent manifold, particularly when dealing with noisy real-world data. Code repository: https://github.com/Thanatorika/mrtg.

Figures

Figures reproduced from arXiv: 2505.04412 by the authors.

Figure 1
Figure 1. (a) An illustration of our model pipeline, (b) an illustration of our manifold reconstruc [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Dimensionality reduction of 3D point cloud data to 2D space. Our proposed method best [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Dimensionality reduction of 100-D spheres, and "rocket" object from PartNet dataset. [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 46 canonical work pages

  1. [8]

    Z. Yao, J. Su, B. Li, and S.-T. Yau. Manifold fitting, 2023. URLhttps://arxiv.org/abs/2304. 07680

    work page 2023

  2. [6]

    Topologicalautoencoders

    M.Moor,M.Horn,B.Rieck,andK.Borgwardt. Topologicalautoencoders. InProceedingsofthe 37th International Conference on Machine Learning, volume 119, pages 7045–7054. PMLR, 13–18 Jul 2020

    work page 2020

  3. [26]

    Y. Lee, S. Yoon, M. Son, and F. C. Park. Regularized autoencoders for isometric represen- tation learning. InInternational Conference on Learning Representations, 2022. URLhttps: //openreview.net/forum?id=mQxt8l7JL04

    work page 2022

  4. [10]

    Measuresofdistortionformachinelearning

    L.C.VankadaraandU.vonLuxburg. Measuresofdistortionformachinelearning. InAdvances in Neural Information Processing Systems, volume 31. Curran Associates, Inc., 2018

    work page 2018

  5. [1]

    Meila and H

    M. Meila and H. Zhang. Manifold learning: what, how, and why.Annual Review of Statistics and Its Application, 11:393–417, 2024

    work page 2024

  6. [2]

    Fefferman, S

    C. Fefferman, S. Mitter, and H. Narayanan. Testing the manifold hypothesis.Journal of the American Mathematical Society, 29:983–1049, 2016

    work page 2016

  7. [3]

    J. B. Tenenbaum, V. de Silva, and J. C. Langford. A global geometric framework for nonlinear dimensionality reduction.Science, 290(5500):2319–2323, 2000

    work page 2000

  8. [4]

    van der Maaten and G

    L. van der Maaten and G. Hinton. Visualizing data using t-sne.Journal of Machine Learning Research, 9(86):2579–2605, 2008. URLhttp://jmlr.org/papers/v9/vandermaaten08a.html

    work page 2008

Show all 51 references
  1. [5]

    McInnes, J

    L. McInnes, J. Healy, N. Saul, and L. Großberger. Umap: Uniform manifold approximation and projection.Journal of Open Source Software, 2018

    2018

  2. [7]

    Goodfellow, Y

    I. Goodfellow, Y. Bengio, and A. Courville.Deep Learning. MIT Press, 2016.http://www. deeplearningbook.org

    2016

  3. [9]

    Persistenthomology—asurvey.Discrete&ComputationalGeom- etry, 453, 01 2008

    H.EdelsbrunnerandJ.Harer. Persistenthomology—asurvey.Discrete&ComputationalGeom- etry, 453, 01 2008

    2008

  4. [11]

    J.-D.Boissonnat,L.J.Guibas,andS.Y.Oudot.Manifoldreconstructioninarbitrarydimensions using witness complexes.Discrete & Computational Geometry, 42:37–70, 2009

    2009

  5. [12]

    Y.-C. Chen, C. R. Genovese, and L. Wassersman. Asymptotic theory for density ridges.The Annals of Statistics, 43:1896–1928, 2015

    1928

  6. [13]

    Lee and B

    D.-T. Lee and B. J. Schachter. Two algorithms for constructing a delaunay triangulation.Inter- national Journal of Computer and Information Sciences, 9:219–242, 1980

    1980

  7. [14]

    Fefferman, S

    C. Fefferman, S. Ivanov, Y. Kurylev, M. Lassas, and H. Narayanan. Fitting a putative manifold to noisy data.Conference On Learning Theory, pages 688–720, 2018

    2018

  8. [15]

    Reconstructionandinterpolation of manifolds

    C.Fefferman,S.Ivanov,M.Lassas,J.Lu,andH.Narayanan. Reconstructionandinterpolation of manifolds. i: The geometric whitney problem.oundations of Computational Mathematic, 20: 1035–1133, 2020

    2020

  9. [16]

    Reconstructionandinterpolation of manifolds ii: Inverse problems for riemannian manifolds with partial distance data, 2021

    C.Fefferman,S.Ivanov,M.Lassas,J.Lu,andH.Narayanan. Reconstructionandinterpolation of manifolds ii: Inverse problems for riemannian manifolds with partial distance data, 2021. URLhttps://arxiv.org/abs/2111.14528

    2021 arXiv

  10. [17]

    Fefferman, S

    C. Fefferman, S. Ivanov, M. Lassas, and H. Narayanan. Fitting a manifold of large reach to noisy data.Journal of Topology and Analysis, 0(0):1–82, 0. doi: 10.1142/S1793525323500012

    doi

  11. [18]

    Genovese, M

    C. Genovese, M. Perone-Pacifico, I. Verdinelli, and L. Wasserman. Minimax manifold estima- tion.Journal of Machine Learning Research, 13:1263–1291, 2012

    2012

  12. [19]

    Genovese, M

    C. Genovese, M. Perone-Pacifico, I. Verdinelli, and L. Wasserman. Manifold estimation and singular deconvolution under hausdorff loss.The Annals of Statistics, 40:941–963, 2012. 10

    2012

  13. [20]

    Psenka, D

    M. Psenka, D. Pai, V. Raman, S Sastry, and Y Ma. Representation learning via manifold flat- tening and reconstruction.Journal of Machine Learning Research, 25(132):1–47, 2024. URL http://jmlr.org/papers/v25/23-0615.html

    2024

  14. [21]

    Representationtopologydivergence: A method for comparing neural network representations

    S.Barannikov,I.Trofimov,N.Balabin,andE.Burnaev. Representationtopologydivergence: A method for comparing neural network representations. InInternational Conference on Machine Learning, 2022

    2022

  15. [22]

    Trofimov, D

    I. Trofimov, D. Cherniavskii, E. Tulchinskii, N. Balabin, S. Barannikov, and E. Burnaev. Learn- ingtopology-preservingdatarepresentations. InInternationalConferenceonLearningRepresen- tations, 2023. URLhttps://openreview.net/forum?id=lIu-ixf-Tzf

    2023

  16. [23]

    Aninvitationtotheeulercharacteristictransform,2023

    E.Munch. Aninvitationtotheeulercharacteristictransform,2023. URLhttps://arxiv.org/ abs/2310.10395

    2023 arXiv

  17. [24]

    Röell and B

    E. Röell and B. Rieck. Differentiable euler characteristic transforms for shape classifica- tion. InThe Twelfth International Conference on Learning Representations, 2024. URLhttps: //openreview.net/forum?id=MO632iPq3I

    2024

  18. [25]

    Geometricautoencoders-whatyouseeiswhat you decode

    N.Philipp,D.Sebastian,andF.A.Hamprecht. Geometricautoencoders-whatyouseeiswhat you decode. InProceedings of the 40th International Conference on Machine Learning, volume 202, pages 25834–25857. PMLR, 23–29 Jul 2023

    2023

  19. [27]

    J. Hahm, J. Lee, S. Kim, and J. Lee. Isometric representation learning for disentangled latent space of diffusion models, 2024. URLhttps://arxiv.org/abs/2407.11451

    2024 arXiv

  20. [28]

    J. M. Lee.Introduction to Smooth Manifolds. Springer, 2012

    2012

  21. [29]

    Federer.Geometric Measure Theory

    H. Federer.Geometric Measure Theory. Springer, 2014

    2014

  22. [30]

    Pedregosa, G

    F. Pedregosa, G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, O. Grisel, M. Blondel, P. Pret- tenhofer, R. Weiss, V. Dubourg, J. Vanderplas, A. Passos, D. Cournapeau, M. Brucher, M. Per- rot, and E. Duchesnay. Scikit-learn: Machine learning in Python.Journal of Machine Learn...

    2011

  23. [31]

    Coenen and A

    A. Coenen and A. Pearce. Understanding umap. URLhttps://pair-code.github.io/ understanding-umap/

  24. [32]

    K. Mo, S. Zhu, A. X. Chang, L. Yi, S. Tripathi, L. J. Guibas, and H. Su. PartNet: A large-scale benchmark for fine-grained and hierarchical part-level 3D object understanding. InThe IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2019

    2019

  25. [33]

    Dan and G

    H. Dan and G. Kevin. Bridging nonlinearities and stochastic regularizers with gaussian error linear units, 2017. URLhttps://openreview.net/forum?id=Bk0MRI5lg

    2017

  26. [34]

    Sainburg, L

    T. Sainburg, L. McInnes, and T. Q. Gentner. Parametric umap embeddings for representation and semisupervised learning.Neural Computation, 33:2881–2907, 2021

    2021

  27. [35]

    InThe 14th European Symposium on Artificial Neural Networks, pages 557–562, 01 2006

    J.VennaandS.Kaski.Visualizinggeneinteractiongraphswithlocalmultidimensionalscaling. InThe 14th European Symposium on Artificial Neural Networks, pages 557–562, 01 2006

    2006

  28. [36]

    Theartofusingt-sneforsinglecelltranscriptomics.NatureCommuni- cations, 10:1–14, 2019

    D.KobakandP.Berens. Theartofusingt-sneforsinglecelltranscriptomics.NatureCommuni- cations, 10:1–14, 2019

    2019

  29. [37]

    U. Bauer. Ripser: efficient computation of vietoris–rips persistence barcodes.Jour- nal of Applicational and Computational Topology, 5, 2021. URLhttps://doi.org/10.1007/ s41468-021-00071-5. 11

    2021

  30. [38]

    stereoscopic

    Richard Liaw, Eric Liang, Robert Nishihara, Philipp Moritz, Joseph E Gonzalez, and Ion Sto- ica. Tune: A research platform for distributed model selection and training.arXiv preprint arXiv:1807.05118, 2018. 12 A. Extended figures and tables FigureS1: Dimensionalityreductionres...

    2018 arXiv

  31. [39]

    Final model

    "Final model" = MRL + AE + TopoReg + GeomReg

  32. [40]

    "Topo AE" = AE + TopoReg

  33. [41]

    "Geom AE" = AE + GeomReg

  34. [42]

    Topo-geom AE

    "Topo-geom AE" = AE + TopoReg + GeomReg

  35. [43]

    MR AE" = MRL + AE. Aboutthe

    "MR AE" = MRL + AE. Aboutthe"mutualpromotioneffect": whenweintroducetheManifoldReconstructionLayer,thedi- mensionalityreductioncomponent(AutoEncoder+tworegularizers)performsbetter,asdepicted in a comparison between "Manifold vs Embedding - Final model" and "Point Cloud vs Embe...

  36. [44]

    Positivity:|v|≤0for allv∈V, with equality if and only ifv= 0

  37. [45]

    Homogeneity:|cv|=|c||v|for allc∈Randv∈V

  38. [46]

    mathematicians cannot tell the difference between a doughnut and a coffee mug

    Triangular inequality:|v+w|≤|v|+|w|for allv,w∈V. A vector space together with a specific choice of norm is called anormed linear space. Example:R n endowed with theEuclidean normdefined by|x|=√x·xis a normed linear space. Definition C2(Support of a function) Iffis any function...

    2000

  39. [47]

    TheKL σ metrics measure the Kullback-Leibler divergence between probability distribu- tionsoftheoriginaldatasetandembedding,wheretheparameterσdefinesalengthscale

  40. [48]

    The kNN metric measures the proportion of nearest neighbors in the embedding that are also nearest neighbors in the original data set [34]

  41. [49]

    The Trust metric is short for trustworthiness, which measures to what extent thek-nearest neighbors of a point are preserved when going from the original dataset to the embedding [35]

  42. [50]

    The RMSE is the root mean square error between the distance matrix of the original data set and the embedding

  43. [51]

    loss"isthedifferencebetweenthecurrentdataandthe noisy input, and the

    The Spear metric measures the Spearman correlation between the distances of all pairs of original and embedding points [36]. Allthemetricscoresareaveragedover10embeddings,whicharerepetitivelytrainedwiththesame hyperparameter settings and input datasets. For kNN and Trust metri...

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.