Pith. sign in

REVIEW 3 major objections 6 minor 66 references

Into the Void: Mapping the Unseen Gaps in High Dimensional Data

T0 review · 3 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read Empty regions of high-dimensional data are reservoirs of novel configurations that a physics-inspired search can find far more reliably than random sampling.

desk verdict Novel L-J-based empty-space search with a solid user study, but the paper's strongest quantitative claim is undermined by a surrogate-as-truth evaluation loop. read the letter →

arxiv 2501.15273 v1 pith:EXERPX77 submitted 2025-01-25 cs.LG cs.HC

Xinyu Zhang , Tyler Estro , Geoff Kuenning , Erez Zadok , Klaus Mueller This is my paper · ORCID
classification cs.LGcs.HC
keywords emptyspacesearchhigh-dimensionaldataLennard-Jonespotentialvisualanalyticshuman-in-the-loopmachinelearningconfigurationoptimizationadversarialreinforcement
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's central assertion is that the holes in a high-dimensional dataset—regions where no verified configuration exists—are not dead space but reservoirs of potentially valuable configurations. It proposes an Empty Space Search Algorithm (ESA) that treats data points as particles and sends repelled agents into these gaps, then wraps the search in a visual-analytics system (GapMiner) and a human-in-the-loop to AI pipeline that trains a neural network to eventually run the search alone. The authors claim this methodology consistently yields substantially better novel configurations than random sampling, and demonstrate it on cache configuration, wine chemistry, adversarial MNIST images, and reinforcement-learning policies. If true, expensive evaluation budgets in high-dimensional design spaces could be spent on a small number of high-value probes instead of large random sweeps.

What carries the argument

The load-bearing object is the Empty Space Search Algorithm (ESA), driven by the Lennard-Jones potential $V(r)=4\epsilon\left[(\sigma/r)^{12}-(\sigma/r)^6\right]$. Each agent sums repulsive and attractive forces from its $k$ nearest data points, moves along the resultant force direction, and returns its trajectory as raw ESCs; a momentum term broadens the search once a DNN can score candidates, and gradient ascent on the DNN refines the final pick. The agent formulation is what keeps the search at $O(dknp + np\log N)$ instead of Delaunay's $O(N^{\lceil d/2\rceil})$, and it is what makes the method parallelizable across many agents.

What would settle it

Run the developed-stage comparison with the true objective instead of the DNN surrogate: take the 1,000- and 3,000-configuration cache datasets, have ESA, random sampling, and random walk each propose 1,500 configurations, then verify the top candidates on the real simulator or trace and compare Pareto dominance areas. If random sampling matches or beats ESA under real verification, the claimed advantage is an artifact of the surrogate's low-density extrapolation.

Watch

Extended reading notes

Core claim

The central discovery is that 'empty space configurations' (ESCs)—points in the data space far from all known verified points—are disproportionately likely to be novel, high-performing, or adversarial, and that a physics-inspired agent search can find them without constructing the Delaunay triangulation whose complexity explodes in high dimensions. ESA places agents in the space, lets Lennard-Jones repulsion and attraction drive them into equilibrium voids, records the trajectory as candidate ESCs, and later uses a trained DNN plus gradient ascent to refine the best candidates. Across a user study and three comparison experiments, the paper reports that ESA significantly outperforms random sampling and random walk in Pareto dominance area, and in the developed stages with a DNN surrogate ESA is much better than random methods.

Load-bearing premise

The developed-stage quantitative evidence assumes the trained DNN (with 7.9% average error on in-distribution data) is trustworthy both for selecting ESCs and for verifying their value; if the surrogate errs systematically in the low-density regions where ESA operates, the measured superiority over random sampling could be an artifact of that extrapolation rather than a real property of the search.

Editorial extensions

If this is right

  • If ESA reliably finds high-value ESCs, design spaces such as storage-cache configurations can be explored with far fewer expensive verifications than random or grid search requires.
  • The DNN-as-critic pipeline suggests that once a surrogate model is accurate enough, the human-analyst role can shift to mentoring rather than direct search, and the same workflow can autocomplete finding adversarial examples in latent spaces.
  • In reinforcement-learning policy-parameter spaces, empty-space search can yield policies whose returns exceed the best training checkpoints, pointing to a less computation-intensive way to improve trained agents.
  • The Pareto dominance-area gains reported in the systems case (from 0.27 to 0.56) imply that the approach can expand the known trade-off frontier rather than merely refine existing designs.

Reading between the lines

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

  • A natural testable extension is to replace the DNN surrogate with the real objective in the developed-stage comparisons; if ESA still beats random sampling, the advantage is a property of the search geometry rather than of surrogate extrapolation.
  • Because ESA's stopping criterion is force magnitude below a threshold, the method effectively defines 'empty' relative to the current dataset; active learning that selects ESCs minimizing DNN uncertainty could make the training loop converge faster.
  • The same empty-space machinery could serve as a data-augmentation generator for imbalanced classification: the wine study already improves quality-prediction accuracy from 60% to 80%, and a similar strategy could probe out-of-distribution examples for robustness testing.
  • The reinforcement-learning result hints at a broader principle: interpolating between nearby high-performing checkpoints in weight space may land on superior policies, a claim that could be checked across other MuJoCo environments.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper presents a visual analytics pipeline, GapMiner, for exploring empty regions in high-dimensional datasets. The central algorithmic contribution is the Empty Space Search Algorithm (ESA), a Lennard-Jones-potential-inspired agent-based search that identifies low-density configurations. The pipeline begins with human-in-the-loop exploration, during which users refine ESA-proposed configurations in GapMiner and verify them against a real (or simulated) black box; the verified configurations augment the dataset and progressively train a DNN that eventually takes over the search and refinement roles. The paper reports a user study (N=10) comparing GapMiner users against ESA-only and random-sampling baselines, two developed-stage comparisons in which ESA is claimed to outperform random sampling and random walk under DNN-guided verification, and three additional case studies: wine dataset augmentation, adversarial images in MNIST latent space, and empty-space policy search in MuJoCo's AntDir. The central claim is that the methodology 'consistently produces substantially superior novel configurations compared to conventional randomization-based methods' (Abstract) and that ESA is 'much better than random methods' (Sec. VIII-C).

Significance. If the central claim were established, the work would be a useful contribution to high-dimensional configuration search and visual analytics: it addresses a real problem (efficient discovery of promising, unverified configurations), and it combines a scalable approximate empty-space search with a human-in-the-loop interface in a way that is more actionable than prior Delaunay-based or flow-line approaches. Genuine strengths include the user study, which uses true verification of the system simulator for the initial-stage comparison and shows a significant advantage for GapMiner users over the baselines; the RL case study, which evaluates discovered policies against actual environment returns; the closed-form derivation connecting PCA updates to original-space coordinate changes (App. A); and the explicit complexity analysis of ESA. However, the quantitative evidence for the developed-stage superiority claim is self-referential: a single DNN surrogate is used both to select candidate ESCs and to compute the outcome values that determine the dominance-area reward. Because ESA deliberately searches low-density or extrapolated regions (App.

major comments (3)
  1. [Sec. VIII-C] The developed-stage comparisons use one well-trained DNN (7.9% average error, measured on in-distribution test data) as both the critic that selects promising ESCs and the black box that verifies their outcome values. Since ESA deliberately searches low-density regions and Appendix E shows that ESA results progressively move away from the initial distribution, the surrogate's accuracy in exactly the regions where ESA operates is not established by its in-distribution error. If the DNN is optimistically biased in those empty regions, the reported dominance-area advantages (0.450 vs. 0.409 for RS and 0.413 for RW; 0.453 vs. 0.418 and 0.422 in the second experiment) could be an artifact of the model's extrapolation behavior rather than a real property of the search. This experiment is the load-bearing evidence for the abstract's claim of 'consistently substantially superior' novel configurations. The authors should validate at least a subset of the developed-stage ESCs against the true simulator (which is already used elsewhere in the paper), or otherwise provide independent evidence of surrogate accuracy in the searched empty regions.
  2. [Sec. IX-A] The wine case study assigns quality labels to newly generated ESC instances as 'the same as the nearest existing neighbor.' The paper then reports that DNN accuracy improved from 60% to 80% and that numerous innovative wine instances with low free sulfur dioxide were found. But the accuracy improvement is measured against these self-generated nearest-neighbor labels, not against real wine quality, and the paper concedes 'we were unable to verify these results in the field.' This is a circular step: the ESA generates points whose labels are defined by nearest-neighbor propagation, and the DNN is trained and evaluated on those same propagated labels. The wine study should be explicitly framed as a proof-of-concept on synthetic labels; it cannot serve as domain validation of the pipeline's ability to find genuinely better wines.
  3. [Sec. VIII-A, Table III] The user study's statistical analysis shows that GapMiner users significantly outperform both baselines, but ESA-only is not significantly different from random sampling (p=0.580 for the x2-x3 pair in the Tukey HSD). This is consistent with the paper's framing of the user as essential in the initial stage, but it undercuts any interpretation of the ESA algorithm itself as superior to random methods without the DNN or user in the loop. The abstract's blanket statement that the methodology 'consistently produces substantially superior novel configurations compared to conventional randomization-based methods' therefore requires the developed-stage evidence to carry the weight; as noted above, that evidence is currently circular. The claim should be scoped to the full pipeline and supported by non-circular validation, or the abstract should be revised.
minor comments (6)
  1. [Sec. IV-B and App. D] ESA has many free parameters (k, sigma, n, alpha, gamma, delta, j), and only a single parameter setting (App. D) is reported. A sensitivity analysis or at least a brief discussion of parameter choice across the different case studies would greatly improve reproducibility.
  2. [Sec. VIII-A] The baseline protocol is described as 'calculated the Pareto front estimated by the naive neural network, and then verified the naive Pareto front.' It should be clarified whether the same (or an equally trained) DNN was available to the GapMiner users during their search; otherwise the comparison may conflate algorithm performance with the quality of the DNN used for candidate selection.
  3. [Sec. V-E and App. B] The cos-MDS neighbor visualization is demonstrated on three synthetic shapes, but no quantitative error metric is given for how well the 2D embedding preserves the original high-dimensional cosine distances. Adding a residual or stress measure would strengthen the claim of faithful topology visualization.
  4. [Sec. VIII-B] The SUS usability score is based on only 5 of the 10 participants, and the paper acknowledges this is an initial test. Reporting the number of responses alongside the score is appropriate, but the limitation should be restated in the main text rather than only implicitly.
  5. [Sec. IX-C] In the RL case study, the empty-space policies are derived from the last 20 checkpoints, and the reported 'Empty-Space Policy' appears to be the best of the 20 candidates. Reporting the distribution of returns across the 20 empty-space policies (rather than only max and mean of one policy's episodes) would help assess selection bias and variance.
  6. [Throughout] There are numerous typographical and formatting artifacts, e.g., 'V oid' in the title, 'T erms' in the index, 'ANVOA' in Appendix D, 'Mujoco' vs. 'MuJoCo', and inconsistent use of 'naive'/'naïve.' These should be cleaned up in a revision.

Circularity Check

2 steps flagged · score 6.0 of 10

Developed-stage ESA-vs-random superiority is measured with the same DNN as selector and verifier, and the wine accuracy gain is measured on self-assigned labels; the user study and RL case study provide independent but narrower evidence.

  1. fitted input called prediction [Section VIII-C (Comparison Experiments)]
    "Based on the advice of field experts, we simplified the pipeline in the block-trace scenario by replacing the critic and the black box with a well-trained DNN. The DNN had an average percentage error of7.9%, working as a surrogate model to simultaneously determine good ESCs and verify outcome values."

    The dominance-area reward is computed from outcome values supplied by the DNN surrogate, while ESA selects and refines ESCs with that same DNN. The 'verified' values are therefore the surrogate's own outputs for both ESA and the random baselines; the experiment measures how well ESA optimizes the surrogate, not real application quality. The in-distribution 7.9% error need not hold in the empty/extrapolated regions ESA targets (Appendix E shows ESA results moving progressively farther from the initial distribution), so a systematic optimistic bias there would be counted as genuine reward and favor ESA. The reported advantage (0.450 vs 0.409/0.413; 0.453 vs 0.418/0.422) is thus unverifiable from the paper alone.

  2. self definitional [Section IX-A (Wine Investigation)]
    "Due to the difficulty of obtaining the quality of new wine instances by ourselves, we made it the same as the nearest existing neighbor. This simple but effective strategy is widely used in imbalanced learning [11]. Having identified over 6,000 ESCs, the DNN accuracy improved from 60% to 80% on quality prediction."

    Augmented wine ESCs receive quality labels copied from the nearest existing neighbor rather than from measurement. The DNN is retrained on these self-assigned labels and its accuracy is computed against the same labels, so the improvement (60% to 80%) measures agreement with the nearest-neighbor assignment, not real wine quality. The paper concedes: 'we were unable to verify these results in the field.' The accuracy gain is therefore a prediction evaluated on its own constructed ground truth.

full rationale

The paper's strongest quantitative claim—that the full pipeline 'consistently produces substantially superior novel configurations compared to conventional randomization-based methods'—rests on the Sec. VIII-C developed-stage comparisons. There, a well-trained DNN (7.9% average in-distribution error) is explicitly used as both the critic that selects promising ESCs and the black box that supplies outcome values; the dominance-area reward is computed from the model's own predictions. This creates a self-referential evaluation loop: ESA is guided by gradient ascent on the same function that later 'verifies' the results, so any optimistic extrapolation in low-density regions is counted as real reward. This is a genuine partial circularity, not merely a correctness concern, because the paper states the substitution verbatim. The wine case study is also circular for accuracy: labels are copied from nearest neighbors and accuracy is reported against those same synthetic labels. However, not all evidence is circular: the user study verifies configurations in the simulator/real-trace setting and shows GapMiner users outperforming ESA and random baselines, and the RL case study evaluates empty-space policies by running them in MuJoCo, giving independent returns. The adversarial-learning MNIST example also uses an independently trained CNN. Therefore the central claim retains independent support in parts, but the developed-stage quantitative superiority claim and the wine accuracy claim reduce to self-referential evaluation; overall score 6.

Assumptions & free parameters 8 free parameters · 6 assumptions · 0 invented entities

The central claim depends on several hand-chosen algorithm parameters, the assumption that empty space corresponds to valuable configurations, the independence of variables, and, most importantly, the assumption that a DNN surrogate can stand in for true evaluation in the main comparison experiments.

free parameters (8)
  • number of neighbors k = 9 in user study and comparison experiments
    Controls the L-J force computation; set to 9 without sensitivity analysis. Chosen by authors.
  • particle effective diameter sigma = mean distance of k neighbors
    Sets the zero-force distance in the L-J potential; data-dependent and recomputed per dataset.
  • search steps n = 400
    Number of L-J integration steps per agent; fixed in experiments.
  • step size alpha = 0.001
    Determines how far the agent moves per step; fixed without sensitivity analysis.
  • momentum discount gamma = 0.9 when DNN is accurate, 0 otherwise
    Controls the momentum term in the agent search; changed as the pipeline evolves.
  • vanishing threshold delta = 1e-7
    Stops the agent when the resultant force magnitude falls below this value.
  • rollout interval j = 10
    Interval at which the agent's trajectory is sampled to form ESCs.
  • DNN error thresholds t1, t2 = 20%, 10% in the system case
    User-specified thresholds; affect when the DNN takes over the search.
assumptions (6)
  • domain assumption Variables are mutually independent
    Explicitly stated in Sec X: 'our ESA assumes that variables are mutually independent.' If false, the Euclidean-distance-based empty space may not correspond to meaningful novel configurations in real causal spaces.
  • domain assumption Empty spaces are reservoirs of valuable novel configurations
    Central premise of the paper, introduced in the Introduction: the paper postulates that unexplored low-density regions contain configurations worth testing. Not proven.
  • ad hoc to paper The DNN surrogate can substitute for true evaluation in developed-stage comparison experiments
    Sec VIII-C: 'we simplified the pipeline ... by replacing the critic and the black box with a well-trained DNN.' All developed-stage reward comparisons rely on this assumption.
  • ad hoc to paper Nearest-neighbor labels for new wine instances are valid ground truth
    Sec IX-A: 'we made it the same as the nearest existing neighbor.' This is used to augment the wine dataset and evaluate the resulting model.
  • domain assumption Lennard-Jones potential models empty space in high-D data
    The algorithm transfers a physics model of intermolecular forces to abstract data space; there is no formal justification that the equilibrium points of the L-J potential coincide with meaningful voids.
  • domain assumption Euclidean distance in raw feature space is the right metric for emptiness
    The ESA uses k-nearest neighbors in Euclidean space, implicitly assuming all features are scaled appropriately and that low Euclidean density is equivalent to 'empty'.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Into the Void: Mapping the Unseen Gaps in High Dimensional Data." pith.science (2026). https://pith.science/paper/EXERPX77

@misc{pith2026250115273,
  author       = {Pith},
  title        = {Pith review of: Into the Void: Mapping the Unseen Gaps in High Dimensional Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EXERPX77}},
  note         = {Machine review of arXiv:2501.15273}
}
read the original abstract

We present a comprehensive pipeline, augmented by a visual analytics system named ``GapMiner'', that is aimed at exploring and exploiting untapped opportunities within the empty areas of high-dimensional datasets. Our approach begins with an initial dataset and then uses a novel Empty Space Search Algorithm (ESA) to identify the center points of these uncharted voids, which are regarded as reservoirs containing potentially valuable novel configurations. Initially, this process is guided by user interactions facilitated by GapMiner. GapMiner visualizes the Empty Space Configurations (ESC) identified by the search within the context of the data, enabling domain experts to explore and adjust ESCs using a linked parallel-coordinate display. These interactions enhance the dataset and contribute to the iterative training of a connected deep neural network (DNN). As the DNN trains, it gradually assumes the task of identifying high-potential ESCs, diminishing the need for direct user involvement. Ultimately, once the DNN achieves adequate accuracy, it autonomously guides the exploration of optimal configurations by predicting performance and refining configurations, using a combination of gradient ascent and improved empty-space searches. Domain users were actively engaged throughout the development of our system. Our findings demonstrate that our methodology consistently produces substantially superior novel configurations compared to conventional randomization-based methods. We illustrate the effectiveness of our method through several case studies addressing various objectives, including parameter optimization, adversarial learning, and reinforcement learning.

Figures

Figures reproduced from arXiv: 2501.15273 by the authors.

Figure 1
Figure 1. Our three-phase workflow. acceleration, which has shown promise in 2D and 3D spaces [7, 47]. However, extending these methods to higher dimensions is non-trivial and this limits their utility in analyzing empty spaces within multivariate datasets. App. C presents empirical studies we conducted that reveals these shortcomings. Other solutions include approximation methods. Peled et al. [26] proposed a Voronoi Diagram… view at source ↗
Figure 2
Figure 2. An example of Lennard-Jones Potential. The [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Given 300 random samples (blue) and 600 random agents (red) in 2D space, [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (14 more)
Figure 4
Figure 4. Figure 4: GapMiner visual interface where a selected ESC is reflected in all displays. (A) Control Panel. From top to bottom: (a) File Selector to load a dataset of initial [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: An example Pareto front at the three key stages. (a) The initial stage when [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Improvement of the Pareto front by interaction in the PCP followed by DNN [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Rewards achieved by users, ESA, and random sampling. Each dot represents [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 9
Figure 9. Figure 9: Empty-space adversarial images from MNIST identified by ESA. We [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 8
Figure 8. Figure 8: The key results of GapMiner on the wine dataset: (a)The PCA map showing [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]
Figure 11
Figure 11. Figure 11: Given 300 random samples (blue) and 600 random agents (red) in 2D [PITH_FULL_IMAGE:figures/full_fig_p016_11.png]
Figure 10
Figure 10. Figure 10: Examples of cos-MDS on 3 synthetic datasets. The blue points are the [PITH_FULL_IMAGE:figures/full_fig_p016_10.png]
Figure 12
Figure 12. Figure 12: Comparison experiment: (a) shows the rewards found by ESA, random [PITH_FULL_IMAGE:figures/full_fig_p017_12.png]
Figure 13
Figure 13. Figure 13: Histograms of the configurations identified by the ESA with regards to [PITH_FULL_IMAGE:figures/full_fig_p018_13.png]
Figure 15
Figure 15. Figure 15: Policy networks visualized by PCA. The color encodes the policy’s return [PITH_FULL_IMAGE:figures/full_fig_p018_15.png]
Figure 16
Figure 16. Figure 16: The full view of GapMiner for the system dataset. The black arrows in the overview display, the PCP, and the Pareto frontier plot show the direction of optimization. [PITH_FULL_IMAGE:figures/full_fig_p019_16.png]
Figure 17
Figure 17. Figure 17: The full view of GapMiner for the wine dataset study. We highlight the global optimal wine instance in green in the PCA map, the PCP, and the neighbor plot. [PITH_FULL_IMAGE:figures/full_fig_p019_17.png]
Figure 18
Figure 18. Figure 18: We trained a neural network to predict forward reward and calculate the control cost; then we calculate the reward of an action as forward reward minus control [PITH_FULL_IMAGE:figures/full_fig_p020_18.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

66 extracted references · 63 canonical work pages

  1. [1]

    Im- age2StyleGAN: How to Embed Images Into the StyleGAN Latent Space?

    Rameen Abdal, Yipeng Qin, and Peter Wonka. “Im- age2StyleGAN: How to Embed Images Into the StyleGAN Latent Space?” In:Proc. ICCV. 2019, pp. 4432–4441

    work page 2019

  2. [2]

    CherryPick: Adaptively unearthing the best cloud configurations for big data analytics

    Omid Alipourfard et al. “CherryPick: Adaptively unearthing the best cloud configurations for big data analytics”. In: USENIX Symposium on Networked Systems Design and Implementation. 2017, pp. 469–482

    work page 2017

  3. [3]

    DeepHull: Fast Convex Hull Approximation in High Dimen- sions

    Randall Balestriero, Zichao Wang, and Richard Baraniuk. “DeepHull: Fast Convex Hull Approximation in High Dimen- sions”. In:ICASSP. 2022, pp. 3888–3892

    work page 2022

  4. [4]

    Determin- ing What Individual SUS Scores Mean: Adding an Adjective Rating Scale

    Aaron Bangor, Philip Kortum, and James Miller. “Determin- ing What Individual SUS Scores Mean: Adding an Adjective Rating Scale”. In:Journal of Usability Studies4.3 (2009), pp. 114–123

    work page 2009

  5. [5]

    Dynamic Programming

    Richard Bellman. “Dynamic Programming”. In:Science 153.3731 (1966), pp. 34–37

    work page 1966

  6. [6]

    Implicit Multidimensional Projection of Local Subspaces

    Rongzheng Bian et al. “Implicit Multidimensional Projection of Local Subspaces”. In:IEEE Transactions on V isualization and Computer Graphics27.2 (2020), pp. 1558–1568

    work page 2020

  7. [7]

    Design and Implementation of a Practical Parallel Delaunay Algorithm

    Guy Blelloch et al. “Design and Implementation of a Practical Parallel Delaunay Algorithm”. In:Algorithmica 24 (1999), pp. 243–269

    work page 1999

  8. [8]

    SUS: A ‘Quick and Dirty’ Usability Scale

    John Brooke. “SUS: A ‘Quick and Dirty’ Usability Scale”. In: Usability Evaluation in Industry189 (Nov. 1995)

    work page 1995

Show all 66 references
  1. [9]

    Data-driven aerospace engineering: reframing the industry with machine learning

    Steven Brunton et al. “Data-driven aerospace engineering: reframing the industry with machine learning”. In:AIAA Journal59.8 (2021), pp. 2820–2847

    2021

  2. [10]

    On Using the Hypervolume Indicator to Compare Pareto Fronts: Applications to Multi-criteria Optimal Experimental Design

    Y ongtao Cao, Byran Smucker, and Timothy Robinson. “On Using the Hypervolume Indicator to Compare Pareto Fronts: Applications to Multi-criteria Optimal Experimental Design”. In: Journal of Statistical Planning and Inference160 (2015), pp. 60–74

    2015

  3. [11]

    SMOTE: Synthetic Minority Over- sampling Technique

    Nitesh Chawla et al. “SMOTE: Synthetic Minority Over- sampling Technique”. In:Journal of Artificial Intelligence Research16 (2002), pp. 321–357

    2002

  4. [12]

    Neural Optimization Machine: A Neural Network Approach for Optimization

    Jie Chen and Y ongming Liu. “Neural Optimization Machine: A Neural Network Approach for Optimization”. In:arXiv preprint arXiv:2208.03897(2022)

    2022 arXiv

  5. [13]

    Distributional- Utility Actor-Critic for Network Slice Performance Guaran- tee

    Jingdi Chen, Tian Lan, and Nakjung Choi. “Distributional- Utility Actor-Critic for Network Slice Performance Guaran- tee”. In:Proc. International Symposium on Theory, Algorith- mic F oundations, and Protocol Design for Mobile Networks and Mobile Computing. 2023, pp. 161–170

    2023

  6. [14]

    Real-Time Network Intrusion De- tection via Decision Transformers

    Jingdi Chen et al. “Real-Time Network Intrusion De- tection via Decision Transformers”. In: arXiv preprint arXiv:2312.07696(2023)

    2023 arXiv

  7. [15]

    The Data Context Map: Fusing Data and Attributes into a Unified Display

    Shenghui Cheng and Klaus Mueller. “The Data Context Map: Fusing Data and Attributes into a Unified Display”. In:IEEE Transactions on V isualization and Computer Graphics22.1 (2015), pp. 121–130

    2015

  8. [16]

    Artificial intelligence and machine learning applications in smart production: Progress, trends, and directions

    Raffaele Cioffi et al. “Artificial intelligence and machine learning applications in smart production: Progress, trends, and directions”. In:Sustainability12.2 (2020), p. 492

    2020

  9. [17]

    Routledge, 2013

    Jacob Cohen et al.Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. Routledge, 2013

    2013

  10. [18]

    Springer Science & Business Media, 2000

    Mark De Berg.Computational Geometry: Algorithms and Applications. Springer Science & Business Media, 2000

    2000

  11. [19]

    Statistical Arbitrage Trading on the Intraday Market Using the Asynchronous Advantage Actor– Critic Method

    Sumeyra Demir et al. “Statistical Arbitrage Trading on the Intraday Market Using the Asynchronous Advantage Actor– Critic Method”. In:Applied Energy314 (2022), p. 118912

    2022

  12. [20]

    A Taxonomy of Clutter Reduction for Information Visualisation

    Geoffrey Ellis and Alan Dix. “A Taxonomy of Clutter Reduction for Information Visualisation”. In:IEEE Transac- tions on V isualization and Computer Graphics13.6 (2007), pp. 1216–1223

    2007

  13. [21]

    Accelerating Multi-Tier Storage Cache Simulations Using Knee Detection

    Tyler Estro et al. “Accelerating Multi-Tier Storage Cache Simulations Using Knee Detection”. In:P erformance Evalu- ation(2024), p. 102410

    2024

  14. [22]

    Desperately Seeking . . . Optimal Multi- Tier Cache Configurations

    Tyler Estro et al. “Desperately Seeking . . . Optimal Multi- Tier Cache Configurations”. In:HotStorage. 2020

    2020

  15. [23]

    Sclow Plots: Visualizing Empty Space

    Joachim Giesen, Lars K¨uhne, and Philipp Lucas. “Sclow Plots: Visualizing Empty Space”. In:Computer Graphics F orum36.3 (2017), pp. 145–155

    2017

  16. [24]

    Approximate Convex Hulls: sketching the convex hull using curvature

    Robert Graham and Adam Oberman. “Approximate Convex Hulls: sketching the convex hull using curvature”. In:arXiv preprint arXiv:1703.01350(2017)

    2017 arXiv

  17. [25]

    Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 13 Stochastic Actor

    Tuomas Haarnoja et al. “Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 13 Stochastic Actor”. In:ICML. 2018, pp. 1861–1870

    2021

  18. [26]

    A Replacement for V oronoi Diagrams of Near Linear Size

    Sariel Har-Peled. “A Replacement for V oronoi Diagrams of Near Linear Size”. In: IEEE Symp. on F oundations of Computer Science. 2001, pp. 94–103

    2001

  19. [27]

    Analysis of a complex of statistical vari- ables into principal components

    Harold Hotelling. “Analysis of a complex of statistical vari- ables into principal components.” In:Journal of Educational Psychology24.6 (1933), p. 417

    1933

  20. [28]

    Auto- matic Algorithm Configuration based on Local Search

    Frank Hutter, Holger H. Hoos, and Thomas St¨utzle. “Auto- matic Algorithm Configuration based on Local Search”. In: AAAI. V ol. 7. 2007, pp. 1152–1157

    2007

  21. [29]

    The plane with parallel coordinates

    Alfred Inselberg. “The plane with parallel coordinates”. In: The V isual Computer1 (1985), pp. 69–91

    1985

  22. [30]

    A Population-Based Hybrid Approach for Hyperparameter Optimization of Neural Networks

    Luis Japa et al. “A Population-Based Hybrid Approach for Hyperparameter Optimization of Neural Networks”. In: IEEE Access11 (2023), pp. 50752–50768.ISSN: 2169-3536. DOI: 10.1109/access.2023.3277310

    2023

  23. [31]

    A Style- Based Generator Architecture for Generative Adversarial Networks

    Tero Karras, Samuli Laine, and Timo Aila. “A Style- Based Generator Architecture for Generative Adversarial Networks”. In:CVPR. 2019, pp. 4401–4410

    2019

  24. [32]

    Causal Reasoning and Large Language Models: Opening a New Frontier for Causality

    Emre Kıcıman et al. “Causal Reasoning and Large Language Models: Opening a New Frontier for Causality”. In:arXiv preprint arXiv:2305.00050(2023)

    2023 arXiv

  25. [33]

    Actor-Critic Algorithms

    Vijay Konda and John Tsitsiklis. “Actor-Critic Algorithms”. In: NeurIPS12 (1999)

    1999

  26. [34]

    Sub- space Clustering

    Hans-Peter Kriegel, Peer Kr¨oger, and Arthur Zimek. “Sub- space Clustering”. In:Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery2.4 (2012), pp. 351–364

    2012

  27. [35]

    On the Determination of Molecular Fields.–—I. From the V ariation of the Viscosity of a Gas with Temperature

    J. Lennard and I. Jones. “On the Determination of Molecular Fields.–—I. From the V ariation of the Viscosity of a Gas with Temperature”. In:Proceedings of the Royal Society of London. Series A106.738 (1924), pp. 441–462

    1924

  28. [36]

    ScatterUQ: Interactive Uncertainty Visual- izations for Multiclass Deep Learning Problems

    Harry Li et al. “ScatterUQ: Interactive Uncertainty Visual- izations for Multiclass Deep Learning Problems”. In:IEEE VIS. 2023, pp. 246–250

    2023

  29. [37]

    Transforming the Latent Space of StyleGAN for Real Face Editing

    Heyi Li et al. “Transforming the Latent Space of StyleGAN for Real Face Editing”. In: The V isual Computer(2023), pp. 1–16

    2023

  30. [38]

    Towards an Information Theoretic Framework of Context-Based Offline Meta-Reinforcement Learning

    Lanqing Li et al. “Towards an Information Theoretic Framework of Context-Based Offline Meta-Reinforcement Learning”. In:arXiv preprint arXiv:2402.02429(2024)

    2024 arXiv

  31. [39]

    Many- to-Many Relational Parallel Coordinates Displays

    Mats Lind, Jimmy Johansson, and Matthew Cooper. “Many- to-Many Relational Parallel Coordinates Displays”. In:IEEE InfoV is. 2009, pp. 25–31

    2009

  32. [40]

    Iterated Local Search

    Helena Lourenc ¸o, Olivier Martin, and Thomas St ¨utzle. “Iterated Local Search”. In:Handbook of Metaheuristics. Springer, 2003, pp. 320–353

    2003

  33. [41]

    Interactive Subspace Cluster Analysis Guided by Semantic Attribute Associa- tions

    Salman Mahmood and Klaus Mueller. “Interactive Subspace Cluster Analysis Guided by Semantic Attribute Associa- tions”. In:IEEE Transactions on V isualization and Computer Graphics(2023)

    2023

  34. [42]

    Illustrative Parallel Coordinates

    Kevin T. McDonnell and Klaus Mueller. “Illustrative Parallel Coordinates”. In:Computer Graphics F orum27.3 (2008), pp. 1031–1038

    2008

  35. [43]

    Umap: Uniform Manifold Approximation and Projection for Di- mension Reduction

    Leland McInnes, John Healy, and James Melville. “Umap: Uniform Manifold Approximation and Projection for Di- mension Reduction”. In:arXiv preprint arXiv:1802.03426 (2018)

    2018 arXiv

  36. [44]

    Review of the Development of Multidimensional Scaling Methods

    Al Mead. “Review of the Development of Multidimensional Scaling Methods”. In:Journal of the Royal Statistical Society: Series D41.1 (1992), pp. 27–39

    1992

  37. [45]

    Human-in-the-loop machine learning: a state of the art

    Eduardo Mosqueira-Rey et al. “Human-in-the-loop machine learning: a state of the art”. In:Artificial Intelligence Review 56.4 (2023), pp. 3005–3054

    2023

  38. [46]

    A Nested Model for Visualization Design and V alidation

    Tamara Munzner. “A Nested Model for Visualization Design and V alidation”. In:IEEE Trans. V isualization and Computer Graphics15.6 (2009), pp. 921–928

    2009

  39. [47]

    Delaunay Triangulation of Large-Scale Datasets Using Two-Level Parallelism

    Cuong Nguyen and Philip Rhodes. “Delaunay Triangulation of Large-Scale Datasets Using Two-Level Parallelism”. In: P arallel Computing98 (2020), p. 102672

    2020

  40. [48]

    An Edge-Bundling Layout for Interactive Parallel Coordinates

    Gregorio Palmas et al. “An Edge-Bundling Layout for Interactive Parallel Coordinates”. In:2014 IEEE P acific V isualization Symposium. IEEE. 2014, pp. 57–64

    2014

  41. [49]

    Provably Efficient Online Hyperparameter Optimization with Population-Based Bandits

    Jack Parker-Holder, Vu Nguyen, and Stephen J Roberts. “Provably Efficient Online Hyperparameter Optimization with Population-Based Bandits”. In:Advances in Neural Information Processing Systems. Ed. by H. Larochelle et al. V ol. 33. Curran Associates, Inc., 2020, pp. 17200–172...

    2020

  42. [50]

    Hubs in Space: Popular Nearest Neighbors in High-Dimensional Data

    Milos Radovanovic, Alexandros Nanopoulos, and Mirjana Ivanovic. “Hubs in Space: Popular Nearest Neighbors in High-Dimensional Data”. In:Journal of Machine Learning Research11 (2010), pp. 2487–2531

    2010

  43. [51]

    Nonlinear Dimensionality Reduction by Locally Linear Embedding

    Sam Roweis and Lawrence Saul. “Nonlinear Dimensionality Reduction by Locally Linear Embedding”. In: Science 290.5500 (2000), pp. 2323–2326

    2000

  44. [52]

    https://www.kaggle.com/ datasets/midouazerty/redwine

    Salah Sammari.red wine data. https://www.kaggle.com/ datasets/midouazerty/redwine. Accessed on 2024-02-01

    2024

  45. [53]

    Measuring Usability LLC, 2011

    Jeff Sauro.A Practical Guide to the System Usability Scale: Background, Benchmarks & Best Practices . Measuring Usability LLC, 2011

    2011

  46. [54]

    The upper bound theorem for polytopes: an easy proof of its asymptotic version

    Raimund Seidel. “The upper bound theorem for polytopes: an easy proof of its asymptotic version”. In:Computational Geometry5.2 (1995), pp. 115–116

    1995

  47. [55]

    Real-Time Visualization of Protein Empty Space with V arying Parameters

    Ondˇrej Strnad et al. “Real-Time Visualization of Protein Empty Space with V arying Parameters”. In: Proc, of BIOTECHNO, IARIA XPS Press(2013), pp. 65–70

    2013

  48. [56]

    ClustNails: Visual analysis of subspace clusters

    Andrada Tatu et al. “ClustNails: Visual analysis of subspace clusters”. In:Tsinghua Science and T echnology17.4 (2012), pp. 419–428

    2012

  49. [57]

    MuJoCo: A physics engine for model-based control

    Emanuel Todorov, Tom Erez, and Y uval Tassa. “MuJoCo: A physics engine for model-based control”. In:IEEE/RSJ International Conference on Intelligent Robots and Systems. 2012, pp. 5026–5033

    2012

  50. [58]

    Visualizing Data Using t-SNE

    Laurens V an der Maaten and Geoffrey Hinton. “Visualizing Data Using t-SNE”. In: Journal of Machine Learning Research9.11 (2008)

    2008

  51. [59]

    Artificial neural networks used in optimization problems

    Gabriel Villarrubia et al. “Artificial neural networks used in optimization problems”. In:Neurocomputing272 (2018), pp. 10–16

    2018

  52. [60]

    Efficient MRC Construction with SHARDS

    Carl Waldspurger et al. “Efficient MRC Construction with SHARDS”. In:F AST. 2015, pp. 95–110

    2015

  53. [61]

    The Subspace V oyager: Exploring High-Dimensional Data along a Continuum of Salient 3D Subspaces

    Bing Wang and Klaus Mueller. “The Subspace V oyager: Exploring High-Dimensional Data along a Continuum of Salient 3D Subspaces”. In:IEEE Transactions on V isualiza- tion and Computer Graphics24.2 (2017), pp. 1204–1222. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 14

    2017

  54. [62]

    High- dimensional data analysis with subspace comparison using matrix visualization

    Junpeng Wang, Xiaotong Liu, and Han-Wei Shen. “High- dimensional data analysis with subspace comparison using matrix visualization”. In: Information V isualization18.1 (2019), pp. 94–109

    2019

  55. [63]

    Learning for personalized medicine: a comprehensive review from a deep learning perspective

    Sushen Zhang et al. “Learning for personalized medicine: a comprehensive review from a deep learning perspective”. In: IEEE Reviews in Biomedical Engineering 12 (2018), pp. 194–208

    2018

  56. [64]

    Graphi- cal Enhancements for Effective Exemplar Identification in Contextual Data Visualizations

    Xinyu Zhang, Shenghui Cheng, and Klaus Mueller. “Graphi- cal Enhancements for Effective Exemplar Identification in Contextual Data Visualizations”. In:IEEE Transactions on V isualization and Computer Graphics(2022)

    2022

  57. [65]

    An Explainable AI Approach to Large Language Model Assisted Causal Model Auditing and Development

    Y anming Zhang et al. “An Explainable AI Approach to Large Language Model Assisted Causal Model Auditing and Development”. In:arXiv preprint arXiv:2312.16211(2023)

    2023 arXiv

  58. [66]

    A Survey on Unsupervised Outlier Detection in High- Dimensional Numerical Data

    Arthur Zimek, Erich Schubert, and Hans-Peter Kriegel. “A Survey on Unsupervised Outlier Detection in High- Dimensional Numerical Data”. In:Statistical Analysis and Data Mining: The ASA Data Science Journal 5.5 (2012), pp. 363–387. Xinyu Zhangearned his B.E. in Computer Science...

    2012

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.