Research map / Learning theory and algorithms / Matrix and tensor methods
Accelerated Alternating Projections for Robust Principal Component Analysis
HanQin Cai, Jian-Feng Cai, Ke Wei
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 accelerated alternating projections (project onto low-dimensional subspace before truncated SVD) instead of AltProj | Robust PCA · Video background subtraction (Shoppingmall) | runtime (s): 82.97s -> 38.98s | Improved |
| use AccAltProj instead of AltProj | Robust PCA · Video background subtraction (Restaurant) | runtime (s): 69.12s -> 28.09s | Improved |
| use AccAltProj instead of GD | Robust PCA · Video background subtraction (Shoppingmall) | runtime (s): 161.1s -> 38.98s | Improved |
| use AccAltProj instead of AltProj for large n | Robust PCA · Synthetic RPCA runtime vs n (r=5, alpha=0.1, c=1) | runtime speedup: about 10x speedup | Improved |
| add trim step to AccAltProj (ablation) | AccAltProj · Video background subtraction (Shoppingmall) | incoherence mu: 2.26 -> 2.12 | Improved |
| add trim step to AccAltProj (ablation) | AccAltProj · Video background subtraction (Restaurant) | incoherence mu: 5.25 -> 5.16 | Improved |
| add trim step to AccAltProj (ablation) | AccAltProj · Video background subtraction (Shoppingmall) | runtime (s): 38.79s -> 38.98s | No clear effect |
| add trim step to AccAltProj (ablation) | AccAltProj · Synthetic RPCA n=2500 r=5 (success counts) | successful recoveries out of 10: identical counts, e.g. 4 vs 4 at c=0.2, alpha=0.7 | No clear effect |
| use AccAltProj instead of GD at small signal level c=0.2 | Robust PCA · Synthetic RPCA n=2500 r=5 c=0.2 | successful recoveries out of 10 at alpha=0.45: 0 -> 10 | Improved |
| use AccAltProj instead of AltProj at alpha=0.65 with c=5 | Robust PCA · Synthetic RPCA n=2500 r=5 c=5 | successful recoveries out of 10 at alpha=0.65: 3 -> 5 | Improved |
| use AccAltProj instead of GD at c=5 | Robust PCA · Synthetic RPCA n=2500 r=5 c=5 | successful recoveries out of 10 at alpha=0.6: 7 -> 10 | Improved |
Closest papers by meaning
- A Nonconvex Projection Method for Robust PCA (AAAI 2019 · similarity 0.90)
- Projected Robust PCA with Application to Smooth Image Recovery (JMLR 2022 · similarity 0.90)
- RES-PCA: A Scalable Approach to Recovering Low-Rank Matrices (CVPR 2019 · similarity 0.89)
- Nearly-Linear Time and Streaming Algorithms for Outlier-Robust PCA (ICML 2023 · similarity 0.89)
- Fast and Provable Nonconvex Tensor RPCA (ICML 2022 · similarity 0.88)
- Efficient Sparse PCA via Block-Diagonalization (ICLR 2025 · similarity 0.86)
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.
