Research map / Learning theory and algorithms / Clustering
k-variates++: more pluses in the 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 Dirac densities with local noise densities and probe functions (k-variates++ general framework) vs k-means++ seeding | k-variates++ · theory: expected potential bound | approximation bound factor | No clear effect |
| Distributed protected seeding (Protected-KM++) vs k-means++ and k-means|| on synthetic data with spread of Forgy nodes | Protected-KM++ · distributed synthetic (m≈20000, d=50) | potential ratio vs k-means++ / k-means|| | Helped |
| Use k-variates++ with identity probes and Laplace noise (sublinear noise regime) vs Forgy-DP | k-variates++ DP · Differential privacy LifeSci/Image/EuropeDiff | relative potential rho' of F-DP (values >1 favor k-means): F-DP ratio 163.0 to 2857.1 | Helped |
| Use k-variates++ DP instead of GUPT | k-variates++ DP · Differential privacy LifeSci (m=26733) | relative potential rho' of GUPT (values <1 favor k-means): GUPT ratio 0.7 on LifeSci, 2.9 on Image, 40.4 on EuropeDiff | Mixed |
| Increase dataset size m under DP (sublinear noise regime) | k-variates++ DP · Differential privacy experiments | relative performance vs baselines | Helped |
Closest papers by meaning
- Fast and Accurate $k$-means++ via Rejection Sampling (NeurIPS 2020 · similarity 0.91)
- Approximate K-Means++ in Sublinear Time (AAAI 2016 · similarity 0.90)
- Differentially Private k-Means via Exponential Mechanism and Max Cover (AAAI 2021 · similarity 0.90)
- Simple, Scalable and Effective Clustering via One-Dimensional Projections (NeurIPS 2023 · similarity 0.90)
- Analyzing $D^\alpha$ seeding for $k$-means (ICML 2024 · similarity 0.89)
- Online Differentially Private Consistent Clustering (NeurIPS 2026 · similarity 0.88)
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.