Research map / Learning theory and algorithms / Matrix and tensor methods
Proximity Operator of the Matrix Perspective Function and its Applications
Joong-Ho (Johann) Won
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 |
|---|---|---|---|
| Newton instead of bisection for root finding | proxphi evaluation · p=1000 | seconds: 1.873 -> 0.8073 | Improved |
| Newton instead of bisection for root finding | proxphi evaluation · p=2000 | seconds: 11.60 -> 4.763 | Improved |
| Newton instead of bisection for root finding | proxphi evaluation · p=500 | iterations: 29.10 -> 7.000 | Improved |
| Newton instead of bisection for root finding | proxphi evaluation · p=50 | KKT: 5.678e-9 -> 1.725-11 | Improved |
| Newton instead of MOSEK SDP solver | proxphi evaluation · p=100 | seconds: 13.60 -> 0.006363 | Improved |
| Newton instead of MOSEK SDP solver | proxphi evaluation · p=30 | KKT: 8.512e-6 -> 1.884e-10 | Improved |
| Bisection instead of MOSEK | proxphi evaluation · p=100 | seconds: 13.60 -> 0.009690 | Improved |
| Increase matrix size p from 10 to 2000 | Newton algorithm · iterations | iterations: 4.900 -> 8.000 | No clear effect |
| Use PDHG with proxphi instead of MOSEK | scaled lasso N=200 · objective | objective: 3.20913 -> 3.20913 | No clear effect |
| Use PDHG on exact pseudolikelihood instead of symmetric lasso | graphical model selection N=30 p=50 · objective | objective: -33.8675 -> -33.9688 | Improved |
| Increase problem size beyond p=500 | PDHG for Gaussian joint MLE · iterations | iterations: 42470 -> 50000 | Hurt |
Closest papers by meaning
- Compact Matrix Factorization With Dependent Subspaces (CVPR 2017 · similarity 0.85)
- A Pairwise Pseudo-likelihood Approach for Matrix Completion with Informative Missingness (NeurIPS 2024 · similarity 0.84)
- Sparse Plus Low Rank Matrix Decomposition: A Discrete Optimization Approach (JMLR 2023 · similarity 0.84)
- Flat Minima in Linear Estimation and an Extended Gauss Markov Theorem (ICLR 2024 · similarity 0.84)
- Spectral Perturbation Bounds for Low-Rank Approximation with Applications to Privacy (NeurIPS 2025 · similarity 0.83)
- Perturbation Bounds for Low-Rank Inverse Approximations under Noise (NeurIPS 2025 · similarity 0.83)
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.
