Research map / Probabilistic and scientific modeling / Variational and generative sampling
Generalized probabilistic principal component analysis of correlated data
Mengyang Gu, Weining Shen
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 |
|---|---|---|---|
| model factor processes as Gaussian processes instead of independent factors (GPPCA vs PCA) | factor loading estimation · Simulation Example 1 (d=4, tau=4, k=8, n=200) | AvgMSE: 1.4×10^-1 -> 5.8×10^-3 | Improved |
| GPPCA instead of PCA at high SNR | factor loading estimation · Simulation Example 1 (d=4, tau=100, k=8, n=200) | AvgMSE: 5.3×10^-3 -> 3.3×10^-4 | Improved |
| GPPCA with different factor covariances instead of PCA | factor loading estimation · Simulation Example 2 (d=4, tau=4, k=8, n=200) | AvgMSE: 1.3×10^-1 -> 1.4×10^-2 | Improved |
| GPPCA instead of PPCA (d=100, fixed noise variance) | PPCA · Gridded temperature interpolation | RMSE: 0.585 -> 0.314 | Improved |
| increase latent factors from 50 to 100 in GPPCA (ablation) | GPPCA (estimated variance) · Gridded temperature interpolation | RMSE: 0.386 -> 0.320 | Improved |
| fix measurement error variance at 0.1 instead of estimating it (ablation) | GPPCA, d=100 · Gridded temperature interpolation | 95% interval coverage: 0.772 -> 0.977 | Improved |
| fix measurement error variance instead of estimating it (ablation) | PPCA, d=100 · Gridded temperature interpolation | 95% interval coverage: 0.525 -> 0.819 | Improved |
| use Matérn kernel instead of Gaussian kernel (ablation) | GPPCA (intercept mean) · DIAMOND simulator emulation | RMSE: 3.33×10^2 -> 2.82×10^2 | Improved |
| add selected covariate (x11) to mean basis (ablation) | GPPCA (Matérn kernel) · DIAMOND simulator emulation | RMSE: 2.82×10^2 -> 2.74×10^2 | Improved |
| add selected covariate to mean basis (ablation) | Ind GP (Gaussian kernel) · DIAMOND simulator emulation | RMSE: 3.64×10^2 -> 4.04×10^2 | Hurt |
| GPPCA instead of independent GPs (Matérn, selected covariates) | Ind GP · DIAMOND simulator emulation | RMSE: 3.31×10^2 -> 2.74×10^2 | Improved |
| GPPCA instead of independent GPs | Ind GP · DIAMOND simulator emulation | 95% interval coverage: 0.927 -> 0.957 | Improved |
Closest papers by meaning
- Correlated-PCA: Principal Components' Analysis when Data and Noise are Correlated (NeurIPS 2016 · similarity 0.89)
- A General Framework for Consistency of Principal Component Analysis (JMLR 2016 · similarity 0.87)
- Probabilistic Geometric Principal Component Analysis with application to neural data (ICLR 2025 · similarity 0.86)
- No Double Descent in Principal Component Regression: A High-Dimensional Analysis (ICML 2024 · similarity 0.85)
- Efficient Sparse PCA via Block-Diagonalization (ICLR 2025 · similarity 0.85)
- Capturing the denoising effect of PCA via compression ratio (NeurIPS 2024 · 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.
