Research map / Probabilistic and scientific modeling / Variational and generative sampling
Shared Subspace Models for Multi-Group Covariance Estimation
Alexander M. Franks, Peter Hoff
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 adaptive shared subspace instead of independent per-group estimation | shared subspace covariance estimator · Simulation, data s=r=2 | Stein's risk: 3.0 -> 0.8 | Improved |
| use adaptive shared subspace instead of pooled single covariance | shared subspace covariance estimator · Simulation, data s=r=2 | Stein's risk: 2.1 -> 0.8 | Improved |
| use adaptive shared subspace instead of independent estimation when covariances are identical | shared subspace covariance estimator · Simulation, identical covariances | Stein's risk: 3.0 -> 0.8 | Improved |
| use adaptive shared subspace instead of pooled covariance when covariances are identical | shared subspace covariance estimator · Simulation, identical covariances | Stein's risk: 0.7 -> 0.8 | No clear effect |
| use adaptive shared subspace instead of pooled covariance when data are full rank | shared subspace covariance estimator · Simulation, s=p=200 | Stein's risk: 138.2 -> 7.1 | Improved |
| use adaptive shared subspace instead of independent estimation when data are full rank | shared subspace covariance estimator · Simulation, s=p=200 | Stein's risk: 3.0 -> 7.1 | Hurt |
| increase fitted shared subspace dimension toward p (ablation) | shared subspace covariance estimator · Simulation, s=p=200, p=200, K=10 | Stein's risk | Improved |
| fit shared subspace of dimension rK rather than smaller when group eigenvectors are dissimilar (ablation) | shared subspace covariance estimator · Simulation, dissimilar eigenvectors | Stein's risk | Improved |
| estimate the shared subspace by maximum marginal likelihood, ignoring its uncertainty | posterior credible regions · Simulation coverage, eigenvalue ratio 3 and 10 | Frequentist coverage of 95% region: 0.94 and 0.92 versus nominal 0.95 | Mixed |
Closest papers by meaning
- Covariance-based Clustering in Multivariate and Functional Data Analysis (JMLR 2016 · similarity 0.89)
- Group-Aware Matrix Estimation and Latent Subspace Recovery (NeurIPS 2026 · similarity 0.86)
- Heterogeneous Sufficient Dimension Reduction and Subspace Clustering (ICML 2025 · similarity 0.85)
- Fundamental Limits of Distributed Covariance Matrix Estimation Under Communication Constraints (ICML 2024 · similarity 0.84)
- A Unified Framework for Variable Selection in Model-Based Clustering with Missing Not at Random (NeurIPS 2025 · similarity 0.84)
- Robust group and simultaneous inferences for high-dimensional single index model (NeurIPS 2024 · similarity 0.84)
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.
