Research map / Learning theory and algorithms / Clustering
Power k-Means Clustering
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 Lloyd's algorithm with power k-means (annealing from s0 toward -infinity via MM) | k-means clustering · Simulated, k=50, d=2 to 200 | Root k-means quality ratio | Helped |
| Replace Lloyd's algorithm with power k-means | k-means clustering · Simulated, d=200 | Variation of information: 0.183 to 0.026 (s0=-3) | Helped |
| Replace KHM (s=-1) with power k-means | k-means clustering · Simulated, d=2 to 200 | Root k-means quality ratio | Helped |
| Initial power s0 set to -18 (rather than -3) for power k-means (ablation) | power k-means · Simulated, d=200 | Root k-means quality ratio: 1.203 vs 1.059 at s0=-3 | Hurt |
| Initial power s0 set to a very low value (-18 or -9) instead of moderate values, so behavior resembles Lloyd's more (ablation) | power k-means · Simulated, d=2 to 200 | Root k-means quality ratio | Mixed |
| Use power k-means on protein expression data instead of k-means++ seeding | k-means clustering of protein data · Protein expression, control and trisomic mice | Quantization error: 0.592 to 0.570 (control); 0.718 to 0.693 (trisomic) | Helped |
| Power k-means used with k-means++ seeding (complementary to seeding) | power k-means · Simulated and real data | Clustering quality | Helped |
Closest papers by meaning
- Simple, Scalable and Effective Clustering via One-Dimensional Projections (NeurIPS 2023 · similarity 0.89)
- The Complexity of k-Means Clustering when Little is Known (ICML 2022 · similarity 0.88)
- Simple and Scalable Sparse k-means Clustering via Feature Ranking (NeurIPS 2020 · similarity 0.88)
- Clustering Semi-Random Mixtures of Gaussians (ICML 2018 · similarity 0.87)
- The Catastrophic Failure of *the* k-Means Algorithm in High Dimensions, and How Hartigan's Algorithm Avoids It (ICML 2026 · similarity 0.87)
- Statistically Optimal $K$-means Clustering via Nonnegative Low-rank Semidefinite Programming (ICLR 2024 · similarity 0.85)
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.