Research map / Learning theory and algorithms / Clustering
Mini-batch $k$-means terminates within $O(d/\epsilon)$ iterations
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 |
|---|---|---|---|
| Terminate mini-batch k-means based on local batch improvement (stop when f_B(C_i) - f_B(C_{i+1}) < epsilon) with batch size b=Omega(kd^3 eps^-2 log) and arbitrary center update rule | generic mini-batch k-means · theoretical bound on iterations | iterations to termination | Helped |
| Use standard update rule with learning rate alpha = sqrt(b_i^j / b) (does not decay to 0) and batch size b=Omega((d/eps)^2 log) instead of kd^3 eps^-2 | mini-batch k-means · theoretical bound on iterations | batch size required for O(d/epsilon) termination | Helped |
| Use sklearn's termination condition (center movement sum below epsilon) with learning rate sqrt(b/b_total) instead of local objective-improvement stopping | sklearn-style mini-batch k-means · theoretical bound on iterations | iterations to termination | Mixed |
| Use k-means++ initialization with mini-batch k-means | mini-batch k-means · solution quality | approximation ratio | No clear effect |
Closest papers by meaning
- Nested Mini-Batch K-Means (NeurIPS 2016 · similarity 0.86)
- Approximate K-Means++ in Sublinear Time (AAAI 2016 · similarity 0.82)
- Near-Linear Time Approximation Algorithms for k-means with Outliers (ICML 2024 · similarity 0.82)
- Analyzing $D^\alpha$ seeding for $k$-means (ICML 2024 · similarity 0.82)
- Relax and Merge: A Simple Yet Effective Framework for Solving Fair $k$-Means and $k$-sparse Wasserstein Barycenter Problems (ICLR 2025 · similarity 0.80)
- Beyond Worst-Case Coreset Bounds for $k$-Clustering via Determinantal Sampling (NeurIPS 2026 · similarity 0.79)
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.