Research map / Probabilistic and scientific modeling / Variational and generative sampling
Fast and Provably Good Seedings for k-Means
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 uniform proposal of K-MC2 with D2-based non-uniform proposal (AFK-MC2 preprocessing) | AFK-MC2 vs K-MC2 · CSN, KDD, SONG (high alpha(X)) | quantization error / distance-evaluation tradeoff | Helped |
| Replace uniform proposal of K-MC2 with non-uniform proposal (AFK-MC2) | AFK-MC2 vs K-MC2 · WEB (low alpha(X)) | quantization error | No clear effect |
| Increase chain length m from 20 to 200 for AFK-MC2 (ablation) | AFK-MC2 · CSN | relative error vs k-means++: 1.45% -> 0.24% | Helped |
| Use AFK-MC2 with small chain length m=20 instead of k-means++ seeding | AFK-MC2 · SUSY, RNA, WEB, KDD, SONG, CSN | speedup in distance evaluations: speedups 13.2x to 1064.7x | Helped |
| Use K-MC2 with chain length m=20 instead of k-means++ | K-MC2 · CSN | relative error vs k-means++: 65.34% error at m=20 | Hurt |
| Uniform random seeding instead of k-means++ seeding | RANDOM baseline · all six datasets | relative error vs k-means++: 4.30% to 915.46% worse | Hurt |
| Increase chain length m for AFK-MC2 (20 to 200) (ablation) | AFK-MC2 · all datasets | quantization error relative to k-means++: error converges toward k-means++ level | Helped |
Closest papers by meaning
- Approximate K-Means++ in Sublinear Time (AAAI 2016 · similarity 0.92)
- Distributed and Provably Good Seedings for k-Means in Constant Rounds (ICML 2017 · similarity 0.91)
- Fast and Accurate $k$-means++ via Rejection Sampling (NeurIPS 2020 · similarity 0.88)
- A Better k-means++ Algorithm via Local Search (ICML 2019 · similarity 0.87)
- Analyzing $D^\alpha$ seeding for $k$-means (ICML 2024 · similarity 0.85)
- Near-Linear Time Approximation Algorithms for k-means with Outliers (ICML 2024 · similarity 0.81)
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.