Research map / Learning theory and algorithms / Matrix and tensor methods
Accelerating Nuclear-norm Regularized Low-rank Matrix Optimization Through Burer-Monteiro Decomposition
Ching-pei Lee, Ling Liang, Tianyun Tang, Kim-Chuan Toh
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 BM-Global instead of QSDPNAL | convex QSDP solving · Regularized kernel estimation, polydism57 | running time (s): 3286.4 -> 305.0 | Improved |
| use BM-Global instead of QSDPNAL | convex QSDP solving · Regularized kernel estimation, polydish57 | running time (s): 3508.9 -> 1765.2 | Improved |
| use BM-Global instead of QSDPNAL | convex QSDP solving · Regularized kernel estimation, WoodyPlants50 | running time (s): 44.3 -> 78.2 | Hurt |
| use BM-Global instead of QSDPNAL | convex QSDP solving · Regularized kernel estimation, delftgestures | running time (s): 318.3 -> 390.1 | Hurt |
| use BM-Global instead of QSDPNAL | convex QSDP solving · Regularized kernel estimation, zongker | optimality eta_opt: 1e-02 -> 1e-04 | Improved |
| use BM-Global instead of QSDPNAL | convex QSDP solving · Molecular conformation, 1LFB | running time (s): 2739.5 -> 25.5 | Improved |
| use BM-Global instead of QSDPNAL | convex QSDP solving · Molecular conformation, 1LFB | RMSD: 1.2 -> 1.5 | Hurt |
| use BM-Global instead of QSDPNAL | convex QSDP solving · Molecular conformation, 1F39 | RMSD: 1.3 -> 0.8 | Improved |
| alternate x=1 inexact proximal gradient step with y=3 BM iterations (ablation) | BM-Global · Matrix completion toy example and movielens100k | objective convergence | No clear effect |
| use BM-Global instead of BM solver alone (polyMF-SS) | matrix factorization · Matrix completion, 4 data sets | relative objective | Improved |
| use BM-Global instead of Active-ALT and AIS-Impute | matrix completion · movielens, netflix, yahoo-music | time to relative RMSE | Improved |
| use fixed step size 1.99 instead of SpaRSA spectral step with linesearch (ablation) | BM-Global · Matrix completion | performance | No clear effect |
Closest papers by meaning
- Fast Hybrid Algorithm for Big Matrix Recovery (AAAI 2016 · similarity 0.87)
- Low-rank Tensor Learning with Nonconvex Overlapped Nuclear Norm Regularization (JMLR 2022 · similarity 0.86)
- Structured Low-Rank Matrix Factorization: Global Optimality, Algorithms, and Applications (TPAMI 2019 · similarity 0.86)
- Sparse Plus Low Rank Matrix Decomposition: A Discrete Optimization Approach (JMLR 2023 · similarity 0.86)
- Large-Scale Low-Rank Matrix Learning with Nonconvex Regularizers (TPAMI 2018 · similarity 0.86)
- Factor Group-Sparse Regularization for Efficient Low-Rank Matrix Recovery (NeurIPS 2019 · 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.
