Research map / Learning theory and algorithms / Kernels and regression theory
PCA of Probability Measures: Sparse and Dense Sampling Regimes
Erell Gachon, Jérémie Bigot, Elsa Cazelles
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 |
|---|---|---|---|
| increase ambient dimension d from 2 to 50 | LOT embedding · Simulated Gaussian measures | covariance risk x100: 2.40 -> 235.64 | Hurt |
| increase ambient dimension d from 2 to 50 | KME embedding · Simulated Gaussian measures | covariance risk x100: 1.43 -> 12.64 | Hurt |
| increase ambient dimension d from 2 to 50 | SW embedding · Simulated Gaussian measures | covariance risk x100: 0.71 -> 0.52 | No clear effect |
| increase ambient dimension d from 2 to 50 | LOT embedding · Simulated Gaussian measures | PCA excess risk x100: 0.07 -> 0.93 | Hurt |
| increase ambient dimension d from 2 to 50 | SW embedding · Simulated Gaussian measures | PCA excess risk x100: 0.02 -> 0.02 | No clear effect |
| increase number of components q from 1 to 10 | SW embedding · Simulated Gaussian measures | PCA excess risk x100: 0.13 -> 7.91 | Hurt |
| increase number of components q from 1 to 10 | KME embedding · Simulated Gaussian measures | PCA excess risk x100: 16.23 -> 53.79 | Hurt |
| use SW instead of KME embedding at q=10 | PCA excess risk · Simulated Gaussian measures | PCA excess risk x100: 53.79 -> 7.91 | Improved |
| increase number of measures n from 10 to 1000 (dense regime) | KME, LOT, SW · Simulated Gaussian measures | covariance risk: slope close to -1/2 | Improved |
| increase number of measures n from 10 to 1000 (dense regime) | KME, LOT, SW · Simulated Gaussian measures | PCA excess risk: slope close to -1 | Improved |
| subsample 50 cells per measure instead of 2000 | PCA with KME, LOT, SW · HIPC flow cytometry (n=63) | 2D PCA representation stability: m=50 vs m=2000 | No clear effect |
Closest papers by meaning
- Combinatorial Sparse PCA Beyond the Spiked Identity Model (ICML 2026 · similarity 0.89)
- Sparse PCA via Covariance Thresholding (JMLR 2016 · similarity 0.88)
- Fast and Provable Algorithms for Sparse PCA with Improved Sample Complexity (ICML 2025 · similarity 0.87)
- Hierarchical Integral Probability Metrics: A distance on random probability measures with low sample complexity (ICML 2024 · similarity 0.85)
- A Geometric Analysis of PCA (NeurIPS 2025 · similarity 0.85)
- Efficient Sparse PCA via Block-Diagonalization (ICLR 2025 · 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.
