Research map / Learning theory and algorithms / Matrix and tensor methods
Informed Non-Convex Robust Principal Component Analysis With Features
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 |
|---|---|---|---|
| Use feature-informed non-convex RPCA instead of the strongest listed baseline | MNIST denoising and classification · MNIST test, α=10–35, linear SVM | accuracy: 86.9 vs 83.35; 84.8 vs 83.4; 83.8 vs 84; 83.15 vs 82.65; 82.05 vs 79; 79.05 vs 73.4 | Mixed |
| Use feature-informed non-convex RPCA instead of the strongest listed baseline | MNIST denoising and classification · MNIST test, α=10–35, RBF SVM | accuracy: 90.3 vs 89.3; 89.8 vs 87.4; 88.55 vs 87.6; 87.8 vs 86.95; 85.65 vs 81.55; 85.15 vs 75.7 | Helped |
| Omit convex projection during factor updates (ablation) | Our algorithm · 200×200 high-incoherence synthetic matrices, ranks 140–155, 10% random-sign errors | recoverability | No clear effect |
| Incorporate feature dictionaries into RPCA | Our algorithm · Synthetic phase-transition recovery | recoverability | Helped |
| Use feature-informed non-convex RPCA rather than PCPF | Our algorithm · Yale B face denoising | visual denoising quality | Helped |
| Use feature-informed non-convex RPCA rather than IRPCA-IHT | Our algorithm · Yale B face denoising | visual denoising quality | Helped |
Closest papers by meaning
- Robust Principal Component Analysis with Side Information (ICML 2016 · similarity 0.92)
- A Pseudo-Bayesian Algorithm for Robust PCA (NeurIPS 2016 · similarity 0.89)
- Outlier-Robust Sparse Estimation via Non-Convex Optimization (NeurIPS 2022 · similarity 0.88)
- Tensor Robust Principal Component Analysis: Exact Recovery of Corrupted Low-Rank Tensors via Convex Optimization (CVPR 2016 · similarity 0.88)
- Rethinking PCA Through Duality (NeurIPS 2025 · similarity 0.86)
- Fast and Provable Algorithms for Sparse PCA with Improved Sample Complexity (ICML 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.