Pith. sign in

REVIEW 4 major objections 4 minor 57 references

When One Point Is Not Enough: Addressing Ambiguous Instances in Dimensionality Reduction by Splitting

T0 review · 4 major / 4 minor · reviewed 2026-08-02 · deepseek-v4-flash

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

Pith's one-line read This paper claims that some data points genuinely belong to multiple dissimilar neighborhoods, that standard dimensionality reduction shows each in only one, and that splitting those points into copies—one per neighborhood—reveals membershi

desk verdict A competent, honest visualization paper with a real artifact and a sound graph-theoretic core, but the load-bearing sparsification-to-LAP step lacks a local guarantee and the 'revelation' is partly by construction. read the letter →

arxiv 2605.23540 v2 pith:3DLXHY6R submitted 2026-05-22 cs.LG

Diede P.M. van der Hoorn , Alessio Arleo , Fernando V. Paulovich This is my paper · ORCID
classification cs.LG
keywords DimensionalityreductionAmbiguousinstancesPartialneighborhoodembeddingLocalarticulationpointsGraphsparsificationVertexsplittingpreservationUMAP
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

The paper identifies a specific, previously unnamed source of distortion in dimensionality reduction: ambiguous instances, points that are highly similar to several mutually dissimilar high-dimensional neighborhoods. Because a standard projection assigns each instance exactly one coordinate, such a point lands in one neighborhood (or none), so the projection silently drops its other memberships—an artifact the authors call partial neighborhood embedding. The paper proposes to detect such instances in the graph that a local DR method builds during its relationship phase, then to split each ambiguous vertex into one copy per true neighborhood before the mapping phase. On image, text, and single-cell examples, the approach places copies of the same instance in each of its neighborhoods, exposing memberships that standard projections hide; existing quality metrics such as trustworthiness and continuity do not flag these cases. If the claim holds, instance-level visual analysis—misclassification audits, rare subpopulation discovery, trajectory interpretation—gains a principled way to see multi-membership structure that is real under the chosen metric.

What carries the argument

The load-bearing object is the r-step local articulation point (LAP): a vertex whose removal disconnects its own r-hop induced subgraph. After effective-resistance spectral sparsification strips away incidental edges, each LAP is treated as ambiguous and split into as many copies as the number of connected components left behind, subject to two filters: edges between two LAPs are dropped, and a component is retained only if its aggregate edge weight reaches a fraction τw of the strongest component. The resulting graph is then embedded by the same local-DR mapping phase, producing one visual copy per true neighborhood.

What would settle it

Take two well-separated clusters in a high-dimensional space and connect them by two adjacent points, each lying between the clusters and similar to both. For any radius r that includes both clusters, removing either bridging point alone leaves the local subgraph connected, so the method reports no ambiguous instances; if ground-truth labels say both points belong to both clusters, partial neighborhood embedding persists despite the method.

Watch

Extended reading notes

Core claim

The central claim is that ambiguity is a structural property of the data-plus-metric, not a layout artifact, and it can be read off the DR relationship graph. After spectral sparsification removes incidental edges, an instance is called ambiguous at radius r when it is a local articulation point: deleting it increases the number of connected components inside its r-step neighborhood. Each such vertex is replaced by as many copies as there are remaining components, with two safeguards—edges linking two ambiguous vertices are removed, and a component is retained only if its total edge weight reaches a user-set fraction τw of the strongest component. The disambiguated graph is then embedded wit

Load-bearing premise

That an ambiguous instance is a single vertex whose removal isolates its true neighborhoods: if two or more borderline points lie between the same neighborhoods, the detector will not fire, so the method misses exactly the graded transitions it highlights.

Editorial extensions

If this is right

  • Analysts auditing a single misclassification can now see both sides: the SVHN '1' misclassified as '7' is placed among sevens and among ones, so the classifier's mistake looks less confident and more explainable.
  • In single-cell visualizations, cells with split copies mark potential transitional or rare intermediate states that a fixed-position projection would have hidden behind one cluster.
  • Because the pipeline operates on the weighted graph produced in the relationship phase, the same detection-and-splitting step carries over to other local graph-based DR techniques, not just the one used for the examples.
  • Trustworthiness and continuity scores are shown to be blind to ambiguity, which motivates a new class of DR quality metrics that accept multiple points per instance.
  • The number of splits as a function of neighborhood size and sparsification is informative: if splits appear only under very aggressive sparsification, the paper suggests ambiguity may not genuinely exist in the data.

Reading between the lines

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

  • Editorial inference: because the definition is metric- and representation-dependent, the same dataset can split under one embedding and not another; that makes the method a diagnostic for whether the chosen representation supports multi-membership, with applications in representation auditing.
  • Editorial inference: the single-vertex LAP criterion cannot see bridges formed by two adjacent borderline points, so in continuous developmental trajectories the most graded transitions are likely the ones the method misses; a trajectory- or pair-aware detector would be a natural next step.
  • Editorial inference: once instances are duplicated, downstream analyses that count points or run clustering on the disambiguated graph need a way to treat copies as soft memberships rather than independent observations; a testable extension is to measure whether soft-membership downstream models improve over hard assignments.
  • Editorial inference: a synthetic benchmark with known ground-truth multi-membership labels—for instance, Gaussian clusters connected by one point versus by a chain—would let users calibrate r and τw and would turn the qualitative examples into an operating-characteristic curve.
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

4 major / 4 minor

Summary. The paper introduces a new visual artifact in dimensionality reduction, 'partial neighborhood embedding,' caused by 'ambiguous instances'—points highly similar to multiple mutually dissimilar high-dimensional neighborhoods. The authors formalize ambiguity as r-step local articulation points (LAPs) on a spectrally sparsified relationship graph (Def. 3.2), split each such vertex into one copy per remaining component (Def. 3.3), and re-embed the modified graph with UMAP. The pipeline is demonstrated on SVHN, MNIST, a text corpus, and single-cell RNA-seq data, with additional experiments on sparsification fidelity and on the insensitivity of standard DR quality metrics. The graph-theoretic core (Lemma 3.1) is simple and correct, but the load-bearing link between spectral sparsification and local LAP detection is only empirically supported by a weak kNN-overlap check.

Significance. If the identified instances genuinely correspond to multi-neighborhood membership, the work gives local DR analysis a principled way to surface structure that standard single-point projections hide. Formalizing ambiguity through graph articulation is novel and the vertex-splitting approach is plausible. The paper also correctly notes that trustworthiness and continuity cannot distinguish ambiguity-driven artifacts from other distortions. The central claim is interesting and potentially useful, but it currently rests on an unvalidated assumption about sparsification preserving local connectivity.

major comments (4)
  1. [Sec. 3.1, Def. 3.2, Eq. (1)] The detector defines ambiguity as LAP in the sparsified graph Gbar, but Eq. (1) is a global Laplacian guarantee. It does not control the number of connected components of the induced r-neighborhood after vertex removal, so sparsification can create or destroy LAPs at radius r. The paper's own parameter choices (ε=0.7–0.9, and the statement that a sparser graph 'allows us to identify more ambiguous instances') make this a practical risk, not a theoretical nicety. The empirical check in Sec. 4.1 (ρHD, Fig. 10) reports kNN overlap of final embeddings, which does not imply equal articulation structure. This gap is load-bearing: if sparsification manufactures LAPs, every 'revealed membership' in Figs. 1, 6–8 may be an artifact. The authors should either prove a local-connectivity transfer lemma or experimentally validate LAP stability across sparsification replicates and against the original
  2. [Sec. 5, Limitations] The paper concedes that 'no two ambiguous instances can exist within the same neighborhood.' This rules out pairs or chains of adjacent borderline points, which are exactly what one expects in developmental single-cell trajectories—the very scenario highlighted in Fig. 8. Consequently, the method cannot detect the most common form of multi-neighborhood membership in continuous processes, and the claim that the approach 'reveals... hidden neighborhood memberships' in such data is unsupported. This limitation is not merely a boundary case; it is a structural restriction of Definition 3.3 and the LAP-LAP deletion rule. The paper should either extend the approach to handle adjacent ambiguities or explicitly narrow its claims to the single-vertex-articulation setting.
  3. [Sec. 3.2, Decision rules] The LAP-LAP rule deletes edges between two LAPs as 'spurious bridges' without formal justification. This presumes that genuine ambiguous vertices are never adjacent, which is the same restrictive assumption noted in Sec. 5. If two vertices form a bridge between neighborhoods (e.g., a differentiation continuum), deleting their connecting edge removes real structure. The heuristic may be reasonable for the examples, but it is not grounded in the formal definition and can erase real multi-neighborhood relationships. Please provide a formal characterization or additional experiments isolating this rule's effect on results.
  4. [Sec. 4.1, Fig. 10] The quantitative sparsification analysis uses only five UMAP runs per graph, and the reported ratios are point estimates without confidence intervals or hypothesis tests. For the RNA-seq dataset, ρ2D is low, but this is attributed to UMAP optimization variance; the same observation could indicate that sparsification changes the layout substantially. More importantly, the analysis never measures the quantity that matters for detection—preservation of the set of LAPs or of component counts in r-neighborhoods—so the conclusion that 'sparsification does not affect local neighborhood structures' is not established at the level needed for the pipeline's core step.
minor comments (4)
  1. [Sec. 3.1, after Def. 3.1] The sentence 'Note that normax can exceed the graph diameter' appears to be a typo; for any vertex, the farthest distance cannot exceed the graph diameter. It should read 'cannot exceed.'
  2. [Sec. 4.1] In the definition of preservedNN@k, the formatting of the equation is broken (missing fraction bar). Please fix the typesetting.
  3. [Sec. 2, Related Work] The comparison to GhostUMAP2 is brief; since both methods produce multiple points per instance, it would help to clarify—beyond 'instability vs. data ambiguity'—whether the two phenomena are distinguishable in practice and whether GhostUMAP2 could also surface ambiguous instances.
  4. [Sec. 3.4 / Sec. 4] The claim of 'near-linear time complexity in practice' is supported only by a heuristic argument and the observed sparsity of LAPs. A brief empirical runtime plot or a more precise complexity bound (e.g., O(n + m + Σ_v |Γ(v)|) under a sparsity assumption) would strengthen the contribution.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: ambiguity is an operational LAP definition and splitting copies into graph components is the mechanism, not a fitted prediction.

full rationale

The load-bearing chain is: build local DR graph G -> sparsify to \bar G -> define LAP_r (Def. 3.1) -> define ambiguity as membership in LAP_r on \bar G (Def. 3.2) -> split each ambiguous vertex into one copy per component of G_r(v)\v (Def. 3.3) -> embed the modified graph. The output memberships are therefore determined by the graph components, but that is the algorithm's construction rather than a hidden reuse of the target claim. The external validity of the revealed memberships is argued from labels/domain knowledge (SVHN true/predicted classes, MNIST pixel differences, ILP/CBR paper topics) that are not used in the detection or split rules, so the examples are not circular. Ref. [34] is a self-citation by two of the authors, but it supplies only the relationship-phase/mapping-phase vocabulary; no theorem from it is needed for Defs. 3.1-3.3. The paper's own limitations (no two ambiguous instances in the same neighborhood; verification of split projections unavailable) are explicit consequences or open problems, not disguised assumptions. The main risk is not circularity but an unproved transfer: Eq. (1) is a global Laplacian guarantee, while LAP detection requires local radius-r connectivity; the empirical rho_HD check (Fig. 10) does not directly certify preservation of articulation structure. That is a correctness/robustness gap, not a reduction of the prediction to the input. Hyperparameters eps and tau_w are tuned per dataset, but the paper does not frame the resulting split counts as an independent quantitative prediction, so this is ordinary parameter selection rather than fitted-input-called-prediction. No equation or definition equates a claimed result with its own input.

Assumptions & free parameters 4 free parameters · 4 assumptions · 2 invented entities

The paper's contribution rests on four hand-set parameters (ε, r, τw, plus external n_neighbors), two domain assumptions that connect graph structure to data semantics, and two ad-hoc definitions that make 'ambiguity' equivalent to the detector's output. The invented concepts (ambiguous instance, PNE) have partial external anchoring via three labeled examples but are otherwise internal to the pipeline. This is typical for an applied visualization paper, but it means the 'reduction of partial neighborhood embedding' claim is not yet independently measurable.

free parameters (4)
  • epsilon (spectral sparsification error) = 0.7 (SVHN, MNIST); 0.9 (text, single-cell)
    Hand-set per dataset; controls how aggressively the relationship graph is pruned and therefore which vertices become LAPs (Secs. 3.1, 4).
  • r (LAP radius) = 2 (all examples)
    Hand-set; 'we use r=2 because it is the most local yet requires instances to be ambiguous on a larger scale' (Sec. 4).
  • tau_w (relative component weight threshold) = 0.05 (SVHN, MNIST, text); 0.1 (single-cell)
    Hand-set per dataset; raised for single-cell because τw=0.05 'resulted in an isolated point' (Sec. 4).
  • UMAP n_neighbors = 15 (SVHN/MNIST), 10 (text), 30 (single-cell)
    External DR parameter, but it determines graph density and hence which vertices can be LAPs (Sec. 4.1, Fig. 9).
assumptions (4)
  • domain assumption The UMAP relationship graph G faithfully encodes true high-dimensional neighborhood structure, so a vertex's cut behavior in G reflects real multi-membership in the data.
    The entire pipeline (Sec. 3) operates on G; the paper notes (Sec. 5) that different techniques or parameterizations yield different ambiguous instances, so the equivalence is technique-relative.
  • domain assumption Spectral sparsification preserves the local radius-r induced-subgraph connectivity needed for LAP detection; the cited guarantee (Eq. 1) covers global Laplacian quadratic forms, not local r-neighborhood cuts.
    Sec. 3.1 invokes the global guarantee; the only support for the local transfer is the empirical ρHD analysis of Sec. 4.1.
  • ad hoc to paper Genuine ambiguity is equivalent to LAP status in the sparsified graph (Def. 3.2), with the number of true neighborhoods equal to the component count after removal (Def. 3.3).
    This operationalization defines the phenomenon by the detector; it inherits the single-vertex blind spot and the τw/LAP-LAP filters.
  • ad hoc to paper LAP-LAP edges are spurious bridges to be deleted, and no two genuine ambiguous instances are adjacent.
    Fig. 5b decision rule; the paper gives no empirical test of how often true ambiguities are adjacent, and Sec. 5 concedes adjacent cases cannot be handled.
invented entities (2)
  • ambiguous instance independent evidence
    purpose: Names the phenomenon of a data point similar to multiple mutually dissimilar neighborhoods and justifies splitting it into copies (Fig. 2).
    Falsifiable handles exist outside the pipeline: the SVHN instance (Fig. 1) is similar to both true and predicted classes under the model's own representation; the MNIST '8' (Fig. 6) is validated by inverse-projected pixel differences. But the construct's definition is parameter-relative (Def. 3.2, ε, r).
  • partial neighborhood embedding (PNE)
    purpose: Names the artifact the method removes; frames the evaluation.
    PNE is defined as 'the loss of neighborhood information for ambiguous instances' (Sec. 1), and Sec. 5 states no current metric can verify it, so the artifact has no independent measurement.

how reviews work

0 comments
Cite this review

Pith. "Pith review of When One Point Is Not Enough: Addressing Ambiguous Instances in Dimensionality Reduction by Splitting." pith.science (2026). https://pith.science/paper/3DLXHY6R

@misc{pith2026260523540,
  author       = {Pith},
  title        = {Pith review of: When One Point Is Not Enough: Addressing Ambiguous Instances in Dimensionality Reduction by Splitting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3DLXHY6R}},
  note         = {Machine review of arXiv:2605.23540}
}
read the original abstract

Dimensionality Reduction (DR) methods are widely used to visualize high-dimensional data. One key task in DR-based analysis is discovering neighborhoods, which relies on analyzing the fine-grained local structure of a projection. However, DR is an inherently lossy process; no technique can perfectly preserve the high-dimensional relationships, and projections therefore contain visual artifacts. In this paper, we highlight a typically overlooked source of visual artifacts: ambiguous instances. These are instances that are highly similar to multiple mutually dissimilar neighborhoods in the high-dimensional space. Standard DR methods cannot faithfully project such instances, since each data instance is mapped to a single point in the visual space. As a result, such an instance is placed in only one of its neighborhoods (or in none at all), so only part of its neighborhood structure is represented. We call this distortion partial neighborhood embedding. In this paper, we introduce a graph-based approach that identifies ambiguous instances and replicates them as multiple points in the projection, placing each copy within its respective neighborhood. We use UMAP for our results, but our approach also generalizes to other local graph-based DR techniques. We show that our approach reveals previously hidden neighborhood memberships in projections and reduces partial neighborhood embedding across multiple examples, and is further supported by quantitative analyses.

Figures

Figures reproduced from arXiv: 2605.23540 by the authors.

Figure 1
Figure 1. We project the latent space of a CNN classifier trained on the Street View House Numbers dataset to study misclassifications. [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. An example of the ambiguous data instance [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Schematic overview of our approach. Given a graph [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Vertex a is identified as LAP at r = 1 and at r = 2 because the neighborhoods are disjoint if a was removed. At r = 3, vertex h connects the previously disjoint neighborhoods. and edges of points b,c,g and e. Removing a creates two separate con￾nected components; {b,c,…
Figure 5
Figure 5. Figure 5: Three scenarios for vertex splitting, showing the resulting graph [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: Projection of the MNIST testset, first reduced to [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: Two projections of the same text dataset containing information on papers, the standard projection suffers from partial neighborhood [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]
Figure 8
Figure 8. Figure 8: Projection of a mice cell landscape, colored by cell-type as in [ [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]
Figure 10
Figure 10. Figure 10: Ratios of G or G¯ embeddings vs high dimensional neighborhood preservation. the alignment between the embedding and the true high-dimensional neighborhoods; we call it ρHD. We used the datasets from the previous examples with the same pa￾rameters described before. We …
Figure 11
Figure 11. Figure 11: Distribution of trustworthiness and continuity for the examples [PITH_FULL_IMAGE:figures/full_fig_p009_11.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

57 extracted references · 4 linked inside Pith

  1. [1]

    M. Aupetit. Visualizing distortions and recovering topology in continuous projection techniques.Neurocomputing, 70(7):1304–1330, Mar. 2007. doi: 10.1016/j.neucom.2006.11.018 1

    doi:10.1016/j.neucom.2006.11.018 2007

  2. [2]

    Barahimi

    F. Barahimi. Multi-point dimensionality reduction to improve projection layout reliability.arXiv preprint arXiv:2101.06224, 2021. doi: 10.48550/ arXiv.2101.06224 2

    arXiv doi:10.48550/arxiv.2101.06224 2021

  3. [3]

    Becht, L

    E. Becht, L. McInnes, J. Healy, C.-A. Dutertre, I. W. H. Kwok, L. G. Ng et al. Dimensionality reduction for visualizing single-cell data using UMAP. Nature Biotechnology, 37(1):38–44, Jan. 2019. doi: 10.1038/nbt.4314 7

    doi:10.1038/nbt.4314 2019

  4. [5]

    Chari and L

    T. Chari and L. Pachter. The specious art of single-cell genomics.PLOS Computational Biology, 19(8):e1011288, Aug. 2023. doi: 10.1371/journal .pcbi.1011288 1

    doi:10.1371/journal 2023

  5. [6]

    Doraiswamy, J

    H. Doraiswamy, J. Tierny, P. J. Silva, L. G. Nonato, and C. Silva. Topomap: A 0-dimensional homology preserving projection of high-dimensional data. IEEE Transactions on Visualization and Computer Graphics, 27(2):561– 571, 2020. doi: 10.1109/TVCG.2020.3030441 2

    arXiv 2020

  6. [7]

    Eppstein, P

    D. Eppstein, P. Kindermann, S. Kobourov, G. Liotta, A. Lubiw, A. Maig- nan et al. On the planar split thickness of graphs.Algorithmica, 80(3):977– 994, 2018. doi: 10.1007/s00453-017-0328-y 3

    doi:10.1007/s00453-017-0328-y 2018

  7. [8]

    Espadoto, R

    M. Espadoto, R. M. Martins, A. Kerren, N. S. Hirata, and A. C. Telea. Toward a quantitative survey of dimension reduction techniques.IEEE Transactions on Visualization and Computer Graphics, 27(3):2153–2173,

  8. [9]

    Feldbauer and A

    R. Feldbauer and A. Flexer. A comprehensive empirical comparison of hubness reduction in high-dimensional spaces.Knowledge and Informa- tion Systems, 59(1):137–166, Apr 2019. doi: 10.1007/s10115-018-1205-y 3

    doi:10.1007/s10115-018-1205-y 2019

Show all 57 references
  1. [10]

    G. S. Green, M. Fujita, H.-S. Yang, M. Taga, A. Cain, C. McCabe et al. Cellular communities reveal trajectories of brain ageing and Alzheimer’s disease.Nature, 633(8030):634–645, Sept. 2024. doi: 10.1038/s41586 -024-07871-6 1

    2024 doi

  2. [11]

    Guardieiro, F

    V . Guardieiro, F. I. De Oliveira, H. Doraiswamy, L. G. Nonato, and C. Silva. Topomap++: A faster and more space efficient technique to compute projections with topological guarantees.IEEE Transactions on Visualization and Computer Graphics, 2024. doi: 10.1109/TVCG.2024. 3456365 2

    2024 doi

  3. [12]

    X. Han, R. Wang, Y . Zhou, L. Fei, H. Sun, S. Lai et al. Mapping the mouse cell atlas by microwell-seq.Cell, 172(5):1091–1107, 2018. 7

    2018

  4. [13]

    Harary.Graph theory (on Demand Printing of 02787)

    F. Harary.Graph theory (on Demand Printing of 02787). CRC Press,

  5. [14]

    Heiter, R

    E. Heiter, R. Vandaele, T. De Bie, Y . Saeys, and J. Lijffijt. Incorporating topological priors into low-dimensional visualizations through topological regularization.IEEE Access, 2024. doi: 10.1109/ACCESS.2024.3456474 2

    2024

  6. [15]

    Henry, A

    N. Henry, A. Bezerianos, and J.-D. Fekete. Improving the readability of clustered social networks using node duplication.IEEE Transactions on Visualization and Computer Graphics, 14(6):1317–1324, 2008. doi: 10. 1109/TVCG.2008.141 3

    2008

  7. [16]

    M. Imre, J. Tao, Y . Wang, Z. Zhao, Z. Feng, and C. Wang. Spectrum- preserving sparsification for visualization of big graphs.Computers & Graphics, 87:89–102, 2020. doi: 10.1016/j.cag.2020.02.004 3

    2020 doi

  8. [17]

    Isenberg, F

    P. Isenberg, F. Heimerl, S. Koch, T. Isenberg, P. Xu, C. Stolper et al. vispubdata.org: A metadata collection about IEEE visualization (VIS) publications.IEEE Transactions on Visualization and Computer Graphics, 23(9):2199–2206, Sept. 2017. doi: 10.1109/TVCG.2016.2615308 9

    2017

  9. [18]

    Jacomy, T

    M. Jacomy, T. Venturini, S. Heymann, and M. Bastian. ForceAtlas2, a continuous graph layout algorithm for handy network visualization designed for the gephi software.PLOS ONE, 9(6):1–12, June 2014. doi: 10.1371/journal.pone.0098679 5

    2014 doi

  10. [19]

    M. Jung, J. Choi, and J. Jo. Projection ensemble: Visualizing the robust structures of multidimensional projections. In2023 IEEE Visualization and Visual Analytics (VIS), pp. 46–50. IEEE, 2023. doi: 10.1109/VIS54172. 2023.00018 2

    2023

  11. [20]

    M. Jung, T. Fujiwara, and J. Jo. Ghostumap2: Measuring and analyzing (r, d)-stability of umap.arXiv preprint arXiv:2507.17174, 2025. doi: 10. 48550/arXiv.2507.17174 2, 9

    2025 doi arXiv

  12. [21]

    Kaski, J

    S. Kaski, J. Nikkilä, M. Oja, J. Venna, P. Törönen, and E. Castrén. Trust- worthiness and metrics in visualizing similarity of gene expression.BMC bioinformatics, 4:1–13, 2003. doi: 10.1186/1471-2105-4-48 2, 8, 9

    2003 doi

  13. [22]

    Kobak and P

    D. Kobak and P. Berens. The art of using t-SNE for single-cell tran- scriptomics.Nature Communications, 10(1):5416, Nov. 2019. doi: 10. 1038/s41467-019-13056-x 1

    2019

  14. [23]

    Lause, P

    J. Lause, P. Berens, and D. Kobak. The art of seeing the elephant in the room: 2d embeddings of single-cell data do make sense.PLOS Computational Biology, 20(10):1–5, 10 2024. doi: 10.1371/journal.pcbi. 1012403 1

    2024 doi

  15. [24]

    LeCun, P

    Y . LeCun, P. Haffner, L. Bottou, and Y . Bengio. Object recognition with gradient-based learning. InShape, contour and grouping in computer vision, pp. 319–345. Springer, 1999. doi: 10.1007/3-540-46805-6_19 7

    1999 doi

  16. [25]

    J. A. Lee and M. Verleysen. Quality assessment of dimensionality reduc- tion: Rank-based criteria.Neurocomputing, 72(7):1431–1443, Mar. 2009. doi: 10.1016/j.neucom.2008.12.017 8

    2009 doi

  17. [26]

    Livi and A

    L. Livi and A. Rizzi. Graph ambiguity.Fuzzy Sets and Systems, 221:24–47, June 2013. doi: 10.1016/j.fss.2013.01.001 2

    2013 doi

  18. [27]

    C. D. Manning, R. Prabhakar, and H. Schütze.Introduction to Information Retrieval. Cambridge University Press, 2008. 7

    2008

  19. [28]

    R. M. Martins, D. B. Coimbra, R. Minghim, and A. Telea. Visual analysis of dimensionality reduction quality for parameterized projections.Com- puters & Graphics, 41:26–42, June 2014. doi: 10.1016/j.cag.2014.01.006 1, 9

    2014 doi

  20. [29]

    R. M. Martins, R. Minghim, and A. C. Telea. Explaining Neighborhood Preservation for Multidimensional Projections. InEG UK Computer Graphics & Visual Computing, 2015. 8

    2015

  21. [30]

    McInnes, J

    L. McInnes, J. Healy, and J. Melville. Umap: Uniform manifold ap- proximation and projection for dimension reduction.arXiv preprint arXiv:1802.03426, 2018. doi: 10.48550/arXiv.1802.03426 2, 6

    2018 doi arXiv

  22. [31]

    Netzer, T

    Y . Netzer, T. Wang, A. Coates, A. Bissacco, B. Wu, A. Y . Ng et al. Reading digits in natural images with unsupervised feature learning. InNIPS workshop on deep learning and unsupervised feature learning, number 2, p. 4. Granada, 2011. 6

    2011

  23. [32]

    Nöllenburg, M

    M. Nöllenburg, M. Sorge, S. Terziadis, A. Villedieu, H.-Y . Wu, and J. Wulms. Planarizing graphs and their drawings by vertex splitting.Jour- nal of Computational Geometry, 16(1):333–372, 2025. doi: 10.20382/jocg .v16i1a10 3

    2025 doi

  24. [33]

    L. G. Nonato and M. Aupetit. Multidimensional projection for visual ana- lytics: Linking techniques with distortions, tasks, and layout enrichment. IEEE Transactions on Visualization and Computer Graphics, 25(8):2650– 2673, 2018. doi: 10.1109/TVCG.2018.2846735 1, 2

    2018

  25. [34]

    F. V . Paulovich, A. Arleo, and S. van den Elzen. When dimensionality re- duction meets graph (drawing) theory: Introducing a common framework, challenges and opportunities.CGF, 44(3):e70105, 2025. doi: 10.1111/cgf .70105 2, 3

    2025 doi

  26. [35]

    F. V . Paulovich, L. G. Nonato, R. Minghim, and H. Levkowitz. Least square projection: A fast high-precision multidimensional projection tech- nique and its application to document mapping.IEEE Transactions on Visualization and Computer Graphics, 14(3):564–575, 2008. doi: 10. ...

    2008

  27. [36]

    Radovanovi´c, A

    M. Radovanovi´c, A. Nanopoulos, and M. Ivanovi´c. Hubs in space: Pop- ular nearest neighbors in high-dimensional data.J. Mach. Learn. Res., 11:2487–2531, Dec. 2010. 3

    2010

  28. [37]

    P. E. Rauber, S. G. Fadel, A. X. Falcão, and A. C. Telea. Visualizing the Hidden Activity of Artificial Neural Networks.IEEE Transactions on Visualization and Computer Graphics, 23(1):101–110, Jan. 2017. doi: 10. 1109/TVCG.2016.2598838 1, 6

    2017

  29. [38]

    P. E. Rauber, A. X. Falcão, and A. C. Telea. Visualizing Time-Dependent Data Using Dynamic t-SNE. InProceedings of the Eurographics/IEEE VGTC Conference on Visualization: Short Papers. Eurographics Associa- tion, 2016. 5

    2016

  30. [39]

    P. C. Ribeiro, G. G. Schardong, S. D. Barbosa, C. S. de Souza, and H. Lopes. Visual exploration of an ensemble of classifiers.Computers & Graphics, 85:23–41, 2019. doi: 10.1016/j.cag.2019.08.012 1

    2019 doi

  31. [40]

    J. Ros, A. Arleo, and F. V . Paulovich. Scalable Force Scheme: A fast method for projecting large datasets.EuroVis Workshop on Visual Analyt- ics (EuroVA) 2025, 2025. doi: 10.2312/EUROV A.20251098 2

    2025 doi

  32. [41]

    S. T. Roweis and L. K. Saul. Nonlinear dimensionality reduction by locally linear embedding.science, 290(5500):2323–2326, 2000. doi: 10. 1126/science.290.5500.2323 2

  33. [42]

    J. W. Sammon. A nonlinear mapping for data structure analysis.IEEE Transactions on computers, 100(5):401–409, 1969. doi: 10.1109/T-C. 1969.222678 2

    1969

  34. [43]

    B. W. B. Shires and C. J. Pickard. Visualizing energy landscapes through manifold learning.Phys. Rev. X, 11:041026, Nov 2021. doi: 10.1103/ PhysRevX.11.041026 1

    2021

  35. [44]

    D. A. Spielman and N. Srivastava. Graph sparsification by effective resistances. InProceedings of the Fortieth Annual ACM Symposium on Theory of Computing, pp. 563–568. ACM, Victoria British Columbia Canada, May 2008. doi: 10.1145/1374376.1374456 3, 4

    2008

  36. [45]

    J. Tang, J. Liu, M. Zhang, and Q. Mei. Visualizing large-scale and high- dimensional data. InProceedings of the 25th international conference on world wide web, pp. 287–297, 2016. doi: 10.1145/2872427.2883041 2

    2016

  37. [46]

    R. Tarjan. Depth-first search and linear graph algorithms.SIAM journal on computing, 1(2):146–160, 1972. doi: 10.1137/0201010 3, 4

    1972 doi

  38. [47]

    Tejada, R

    E. Tejada, R. Minghim, and L. G. Nonato. On improved projection techniques to support visual exploration of multi-dimensional data sets. Information Visualization, 2(4):218–231, 2003. doi: 10.1057/palgrave.ivs. 9500054 2

    2003 doi

  39. [48]

    J. B. Tenenbaum, V . d. Silva, and J. C. Langford. A global geometric frame- work for nonlinear dimensionality reduction.science, 290(5500):2319– 2323, 2000. doi: 10.1126/science.290.5500.2319 2

  40. [49]

    W. S. Torgerson. Multidimensional scaling: I. theory and method.Psy- chometrika, 17(4):401–419, 1952. doi: 10.1007/BF02288916 2

    1952 doi

  41. [50]

    Van der Maaten and G

    L. Van der Maaten and G. Hinton. Visualizing data using t-sne.Journal of machine learning research, 9(11), 2008. 2, 7

    2008

  42. [51]

    Van der Maaten and G

    L. Van der Maaten and G. Hinton. Visualizing non-metric similarities in multiple maps.Machine learning, 87(1):33–55, 2012. doi: 10.1007/ s10994-011-5273-4 2

    2012

  43. [52]

    R. G. Viegas, I. B. S. Martins, M. N. Sanches, A. B. Oliveira Junior, J. B. d. Camargo, F. V . Paulovich et al. Elvim: Exploring biomolecular energy landscapes through multidimensional visualization.Journal of Chemical Information and Modeling, 64(8):3443–3450, Apr 2024. doi: ...

    2024 doi

  44. [53]

    F. Wan, M. D. T. Torres, J. Peng, and C. de la Fuente-Nunez. Deep- learning-enabled antibiotic discovery through molecular de-extinction. Nature Biomedical Engineering, 8(7):854–871, July 2024. doi: 10.1038/ s41551-024-01201-x 1

    2024

  45. [54]

    Y . Wang, Q. Shen, D. Archambault, Z. Zhou, M. Zhu, S. Yang et al. Ambi- guityvis: Visualization of ambiguity in graph layouts.IEEE Transactions on Visualization and Computer Graphics, 22(1):359–368, 2016. doi: 10. 1109/TVCG.2015.2467691 3

    2016

  46. [55]

    Yan and W

    K. Yan and W. Cui. Visualizing the uncertainty induced by graph layout algorithms. In2017 IEEE Pacific Visualization Symposium (PacificVis), pp. 200–209, 2017. doi: 10.1109/PACIFICVIS.2017.8031595 3

    2017

  47. [56]

    S. Yu, L. Gao, and R. Song. Local articulation points in complex networks. arXiv preprint arXiv:1812.10631, 2018. doi: 10.48550/arXiv.1812.10631 4

    2018 doi arXiv

  48. [2018]

    doi: 10.1201/9780429493768 3

    doi

  49. [2019]

    doi: 10.1109/TVCG.2019.2944182 1, 2

    2019

Pith tools

Reviewed August 2, 2026 · model on record in the stance chip above.