Research map / Learning theory and algorithms / Matrix and tensor methods
Dual Principal Component Pursuit for Robust Subspace Learning: Theory and Algorithms for a Holistic Approach
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 |
|---|---|---|---|
| holistic DPCP (simultaneous Grassmannian basis) instead of recursive single-normal DPCP | DPCP for codimension-c subspace · Synthetic D=1000, c=50, N=10D | running time: 1.78s vs 79.74s | Helped |
| holistic DPCP instead of recursive DPCP | DPCP for codimension-c subspace · Synthetic D=1000, c=50, N=10D, high noise and outlier ratio | subspace estimation accuracy | Helped |
| PCA and REAPER baselines instead of holistic DPCP | robust subspace recovery · Synthetic D=1000, c=50, N=10D | subspace estimation accuracy | Hurt |
| holistic DPCP performs similarly to GGD and GGD-dual in accuracy but runs much faster | robust subspace recovery · Synthetic D=1000, c=50, N=10D | running time: 1.78s vs 24.78s (GGD) | Helped |
| noise level sigma from 0 to 1e-6 effect on RSGM convergence | holistic DPCP with RSGM · Synthetic D=30, N=500, outlier ratio 0.7 | convergence to S-perp neighborhood | No clear effect |
| holistic DPCP tolerates O(N^2) outliers (theory) vs O(N) for other RSR methods | robust subspace recovery · theoretical analysis, noiseless and noisy | outlier tolerance | Helped |
Closest papers by meaning
- Implicit Bias of Projected Subgradient Method Gives Provable Robust Recovery of Subspaces of Unknown Codimension (ICLR 2022 · similarity 0.94)
- Dual Principal Component Pursuit: Improved Analysis and Efficient Algorithms (NeurIPS 2018 · similarity 0.93)
- Noisy Dual Principal Component Pursuit (ICML 2019 · similarity 0.91)
- A Linearly Convergent Method for Non-Smooth Non-Convex Optimization on the Grassmannian with Applications to Robust Subspace and Dictionary Learning (NeurIPS 2019 · similarity 0.91)
- Hyperplane Clustering via Dual Principal Component Pursuit (ICML 2017 · similarity 0.89)
- Fun with Flags: Robust Principal Directions via Flag Manifolds (CVPR 2024 · similarity 0.87)
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.