Toward a Quantitative Survey of Dimension Reduction Techniques

Espadoto, Mateus; Martins, Rafael Messias; Kerren, Andreas; Hirata, Nina S. T.; Telea, Alexandru

doi:10.1109/tvcg.2019.2944182

Cited by 236 publications

(296 citation statements)

References 71 publications

Supporting

Mentioning

292

Contrasting

Order By: Relevance

“…They are shown as a grayscale heatmap under each cell of the thumbnail matrix ( Figure 2). For more details on the quality measures, please refer to [9]. It is important to clarify, however, that these quality measures are offered only as a support for the visual analysis.…”

Section: Goal 1: Hyper-parameter Explorationmentioning

confidence: 99%

t-viSNE: Interactive Assessment and Interpretation of t-SNE Projections

Chatzimparmpas

Martins

Kerren

2020

IEEE Trans. Visual. Comput. Graphics

Self Cite

127

View full text Add to dashboard Cite

t-Distributed Stochastic Neighbor Embedding (t-SNE) for the visualization of multidimensional data has proven to be a popular approach, with successful applications in a wide range of domains. Despite their usefulness, t-SNE projections can be hard to interpret or even misleading, which hurts the trustworthiness of the results. Understanding the details of t-SNE itself and the reasons behind specific patterns in its output may be a daunting task, especially for non-experts in dimensionality reduction. In this work, we present t-viSNE, an interactive tool for the visual exploration of t-SNE projections that enables analysts to inspect different aspects of their accuracy and meaning, such as the effects of hyper-parameters, distance and neighborhood preservation, densities and costs of specific neighborhoods, and the correlations between dimensions and visual patterns. We propose a coherent, accessible, and well-integrated collection of different views for the visualization of t-SNE projections. The applicability and usability of t-viSNE are demonstrated through hypothetical usage scenarios with real data sets. Finally, we present the results of a user study where the tool's effectiveness was evaluated. By bringing to light information that would normally be lost after running t-SNE, we hope to support analysts in using t-SNE and making its results better understandable.

show abstract

Section: Goal 1: Hyper-parameter Explorationmentioning

confidence: 99%

t-viSNE: Interactive Assessment and Interpretation of t-SNE Projections

Chatzimparmpas

Martins

Kerren

2020

IEEE Trans. Visual. Comput. Graphics

Self Cite

127

View full text Add to dashboard Cite

show abstract

“…Rather, for choosing which dimensionality reduction we will next use in our work, we will consider a separate recent survey [37]. In this survey, the authors analyze 44 actual implementations of projection techniques against 20 datasets, using six di erent quality metrics from the literature.…”

Section: Dimensionality Reductionmentioning

confidence: 99%

“…The method is computationally less expensive than t-SNE, has easy-to-set parameters and, as already mentioned, has the out-of-sample capability, which is important when one wants to project the same (or related) dataset(s) multiple times and compare the projections. A more detailed comparison of t-SNE and UMAP is given in [37,87]. Given all above, we also consider UMAP, along with LAMP and t-SNE, in our work next.…”

Section: Umap: Uniform Manifold Approximation and Projectionmentioning

confidence: 99%

“…The "design space" of all possible combinations for constructing decision boundary maps is very large, as it involves direct projections, inverse projections, classi er techniques, actual datasets, and hyperparameter values. In related work, Espadoto et al [37] have attempted to analyze a related, but much lower-dimensional, space consisting of the behavior of projection techniques for di erent types of datasets and hyperparameter values, and for this end they needed to evaluate thousands of combinations. Given that our design space has more dimensions, we will not attempt to densely sample it for evaluation.…”

Section: E Va L U At I N G D E C I S I O N B O U N D a R Y M A P Smentioning

confidence: 99%

“…Which ones are best for constructing decision boundary maps is, however, not evident. To nd these, we can use, up to some extent, the analysis of Espadoto et al [37] that quantitatively compared over 40 such techniques against six di erent quality metrics. However, the respective quality metrics are chie y aimed at gauging how well a projection succeeds in creating a scatterplot that faithfully conveys the highdimensional data structure.…”

Section: E Va L U At I N G D E C I S I O N B O U N D a R Y M A P Smentioning

confidence: 99%

See 2 more Smart Citations

Visual Analytics for Machine Learning: Computing and Leveraging Decision Boundary Maps

Rodrigues¹

View full text Add to dashboard Cite

show abstract

From High Dimensions to Human Insight: Exploring Dimensionality Reduction for Chemical Space Visualization

Orlov,

Akhmetshin,

Horvath

et al. 2024

Molecular Informatics

View full text Add to dashboard Cite

Dimensionality reduction is an important exploratory data analysis method that allows high‐dimensional data to be represented in a human‐interpretable lower‐dimensional space. It is extensively applied in the analysis of chemical libraries, where chemical structure data ‐ represented as high‐dimensional feature vectors‐are transformed into 2D or 3D chemical space maps. In this paper, commonly used dimensionality reduction techniques ‐ Principal Component Analysis (PCA), t‐Distributed Stochastic Neighbor Embedding (t‐SNE), Uniform Manifold Approximation and Projection (UMAP), and Generative Topographic Mapping (GTM) ‐ are evaluated in terms of neighborhood preservation and visualization capability of sets of small molecules from the ChEMBL database.

show abstract

Toward a Quantitative Survey of Dimension Reduction Techniques

Cited by 236 publications

References 71 publications

t-viSNE: Interactive Assessment and Interpretation of t-SNE Projections

t-viSNE: Interactive Assessment and Interpretation of t-SNE Projections

Visual Analytics for Machine Learning: Computing and Leveraging Decision Boundary Maps

From High Dimensions to Human Insight: Exploring Dimensionality Reduction for Chemical Space Visualization

Contact Info

Product

Resources

About