Research map / Probabilistic and scientific modeling / Molecules and proteins
Understanding How Dimension Reduction Tools Work: An Empirical Approach to Deciphering t-SNE, UMAP, TriMap, and PaCMAP for Data Visualization
Yingfan Wang, Haiyang Huang, Cynthia Rudin, Yaron Shaposhnik
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use PaCMAP instead of t-SNE | t-SNE · Synthetic hierarchical dataset (50-d, 125 micro clusters) | Random triplet accuracy: 0.508 ± 0.003 -> 0.801 ± 0.002 | Improved |
| use PaCMAP instead of UMAP | UMAP · Synthetic hierarchical dataset (50-d, 125 micro clusters) | Random triplet accuracy: 0.506 ± 0.001 -> 0.801 ± 0.002 | Improved |
| use PaCMAP instead of TriMap | TriMap · Synthetic hierarchical dataset (50-d, 125 micro clusters) | Random triplet accuracy: 0.665 ± 0.003 -> 0.801 ± 0.002 | Improved |
| use PaCMAP instead of LargeVis | LargeVis · Synthetic hierarchical dataset (50-d, 125 micro clusters) | Random triplet accuracy: 0.506 ± 0.004 -> 0.801 ± 0.002 | Improved |
| use PaCMAP instead of t-SNE | t-SNE · Synthetic hierarchical dataset (50-d, 125 micro clusters) | Centroid triplet accuracy: 0.499 ± 0.002 -> 0.794 ± 0.001 | Improved |
| use PaCMAP instead of UMAP | UMAP · Synthetic hierarchical dataset (50-d, 125 micro clusters) | Centroid triplet accuracy: 0.497 ± 0.001 -> 0.794 ± 0.001 | Improved |
| use PaCMAP instead of TriMap | TriMap · Synthetic hierarchical dataset (50-d, 125 micro clusters) | Centroid triplet accuracy: 0.651 ± 0.003 -> 0.794 ± 0.001 | Improved |
| use PaCMAP instead of LargeVis | LargeVis · Synthetic hierarchical dataset (50-d, 125 micro clusters) | Centroid triplet accuracy: 0.496 ± 0.006 -> 0.794 ± 0.001 | Improved |
| use PaCMAP instead of t-SNE | t-SNE · Synthetic hierarchical dataset (50-d, 125 micro clusters) | 1NN accuracy: 1.000 ± 0.000 -> 1.000 ± 0.000 | No clear effect |
| use PaCMAP instead of UMAP | UMAP · Synthetic hierarchical dataset (50-d, 125 micro clusters) | SVM accuracy: 1.000 ± 0.000 -> 1.000 ± 0.000 | No clear effect |
| use PaCMAP instead of t-SNE (running time) | t-SNE · Synthetic hierarchical dataset (50-d, 125 micro clusters) | running time: 00:02:18 -> 00:00:24 | Improved |
| use PaCMAP instead of UMAP (running time) | UMAP · Synthetic hierarchical dataset (50-d, 125 micro clusters) | running time: 00:00:45 -> 00:00:24 | Improved |
Closest papers by meaning
- Dimension Reduction with Locally Adjusted Graphs (AAAI 2025 · similarity 0.87)
- Partition First, Embed Later: Laplacian-Based Feature Partitioning for Refined Embedding and Visualization of High-Dimensional Data (ICML 2025 · similarity 0.87)
- Finsler Multi-Dimensional Scaling: Manifold Learning for Asymmetric Dimensionality Reduction and Embedding (CVPR 2025 · similarity 0.86)
- Navigating the Effect of Parametrization for Dimensionality Reduction (NeurIPS 2024 · similarity 0.86)
- Generalized Dimension Reduction Using Semi-Relaxed Gromov-Wasserstein Distance (AAAI 2025 · similarity 0.85)
- Neuc-MDS: Non-Euclidean Multidimensional Scaling Through Bilinear Forms (NeurIPS 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
