Research map / Learning theory and algorithms / Clustering
Selective inference for k-means clustering
Yiqun T. Chen, Daniela M. Witten
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 |
|---|---|---|---|
| use selective p-value instead of naive Wald p-value on no-cluster scRNA-seq data | k-means difference-in-means test · scRNA-seq no-cluster memory T cells | p-value / Type I error: <10^{-10} -> 0.30 | Improved |
| use selective p-value instead of naive p-value under global null | k-means difference-in-means test · simulation global null, n=150, q=2,10,50,100 | selective Type I error | Improved |
| use selective p-value instead of naive on MNIST zeros with no clusters | k-means difference-in-means test · MNIST no-cluster (1,500 zeros) | p-value | Improved |
| use selective p-value instead of naive on penguins | k-means difference-in-means test · Palmer Penguins (165 females, K=4) | p-value | Improved |
| use known-variance sigma-free MED estimator instead of sample-variance estimator (ablation) | selective p-value · simulation conditional power, n=150, q=10 | conditional power | Improved |
| increase distance delta between true clusters | selective p-value · simulation, q=10, delta=2..10 | detection probability and conditional power | Improved |
| increase noise sigma (0.25, 0.5, 1) | selective p-value · simulation, q=10 | conditional power | Hurt |
| sample splitting with k-NN labelling of test set | naive Wald test · simulation null, 100x2 | Type I error | No clear effect |
| apply selective test on data with true clusters | selective p-value · scRNA-seq cluster dataset (5 cell types) | p-value: 4.0\times 10^{-4} | Improved |
Closest papers by meaning
- Clustering Small Samples With Quality Guarantees: Adaptivity With One2all PPS (AAAI 2018 · similarity 0.86)
- The Impact of Random Models on Clustering Similarity (JMLR 2017 · similarity 0.84)
- Improved Learning-augmented Algorithms for k-means and k-medians Clustering (ICLR 2023 · similarity 0.84)
- Detecting Meaningful Clusters From High-Dimensional Data: A Strongly Consistent Sparse Center-Based Clustering Approach (TPAMI 2020 · similarity 0.84)
- Data-Efficient Learning via Clustering-Based Sensitivity Sampling: Foundation Models and Beyond (ICML 2024 · similarity 0.84)
- Statistical Test for Feature Selection Pipelines by Selective Inference (ICML 2025 · similarity 0.83)
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.
