Research map / Learning theory and algorithms / Clustering
On Perfect Clustering of High Dimension, Low Sample Size Data
Soham Sarkar, Anil Kumar Ghosh
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 |
|---|---|---|---|
| replace Euclidean distance with MADD dissimilarity | average linkage · Ex-3 (d=100) | Rand index (lower is better): 0.6616 -> 0.0000 | Improved |
| replace Euclidean distance with MADD dissimilarity | k-means · Ex-2 (d=500) | Rand index (lower is better): 0.6982 -> 0.0000 | Improved |
| replace RBF-of-Euclidean similarity with MADD-based similarity | spectral clustering · Ex-6 (d=100) | Rand index (lower is better): 0.5000 -> 0.0000 | Improved |
| replace Euclidean distance with MADD dissimilarity | average linkage · Ex-1 (d=100) | Rand index (lower is better): 0.2906 -> 0.0865 | Improved |
| replace Euclidean distance with MADD dissimilarity | k-means · Ex-1 (d=100) | Rand index (lower is better): 0.0185 -> 0.0367 | Hurt |
| use rho2 (bounded psi) instead of rho0 (ablation) | k-means with MADD · Ex-7 (d=100) | Rand index (lower is better): 0.4102 -> 0.0001 | Improved |
| use rho2 (bounded psi) instead of rho0 (ablation) | average linkage with MADD · Ex-8 (d=100) | Rand index (lower is better): 0.5020 -> 0.1309 | Improved |
| use rho1 instead of rho0 (ablation) | average linkage with MADD · Ex-7 (d=500) | Rand index (lower is better): 0.4776 -> 0.0471 | Improved |
| use rho2 (bounded psi) instead of rho1 (identity psi) (ablation) | k-means with MADD · Ex-8 (d=100) | Rand index (lower is better): 0.3132 -> 0.0845 | Improved |
| use MADD with bounded psi when t3 is replaced by Cauchy | MADD clustering · Ex-8 with Cauchy | Rand index (lower is better): close to 0 versus close to 0.5 | Improved |
| use MADD with rho0 in the Dunn/PD estimators of cluster number | MADD-based PD estimator · Ex-7 and Ex-8 | number of clusters estimated | Hurt |
| use rho1 or rho2 instead of rho0 in the PD estimator (ablation) | MADD-based PD estimator · Ex-7 and Ex-8 | number of clusters estimated | Improved |
Closest papers by meaning
- A Sampling-Based Approach for Efficient Clustering in Large Datasets (CVPR 2022 · similarity 0.88)
- Detecting Meaningful Clusters From High-Dimensional Data: A Strongly Consistent Sparse Center-Based Clustering Approach (TPAMI 2020 · similarity 0.86)
- Simple, Scalable and Effective Clustering via One-Dimensional Projections (NeurIPS 2023 · similarity 0.86)
- Subquadratic High-Dimensional Hierarchical Clustering (NeurIPS 2019 · similarity 0.85)
- Uniform Concentration Bounds toward a Unified Framework for Robust Clustering (NeurIPS 2021 · similarity 0.85)
- Beyond Worst-Case Coreset Bounds for $k$-Clustering via Determinantal Sampling (NeurIPS 2026 · 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.
