Research map / Learning theory and algorithms / Matrix and tensor methods
Fast Projected Newton-like Method for Precision Matrix Estimation under Total Positivity
Jian-Feng CAI, José Vinícius De Miranda Cardoso, Daniel P. Palomar, Jiaxi Ying
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 FPN's approximate Newton direction instead of the exact Newton direction (ablation) | FPN · BA graph synthetic | objective convergence | Improved |
| FPN versus BCD, GGL, optGL, PGD, APGD, PPA, PQN-LBFGS | FPN · BA degree one, p 1000 to 5000 | convergence time | Improved |
| FPN versus baselines on degree-two BA graphs | FPN · BA degree two | convergence time | Improved |
| FPN with disconnectivity constraints versus baselines | FPN · BA graphs with disconnectivity | convergence time | Improved |
| increase node count to p=5000 | BCD and GGL · BA graph | convergence time | Hurt |
| use second-order information (FPN, PQN-LBFGS) instead of first-order PGD/APGD | FPN, PQN-LBFGS · BA graph p=5000 | iterations to converge | Improved |
| vary sparsity level via regularization parameter (ablation) | BCD, GGL, PPA, FPN · BA graph | run time | No clear effect |
| decrease sparsity of estimate (ablation) | PGD, APGD, PQN-LBFGS · BA graph | run time | Hurt |
| FPN versus baselines on concepts dataset (p=1000) | FPN · concepts | convergence time | Improved |
| MTP2 FPN instead of Glasso | FPN · S&P 500 stocks | modularity: 0.47 -> 0.65 | Improved |
| add disconnectivity set to FPN (ablation) | FPN · S&P 500 stocks | modularity: 0.65 -> 0.67 | Improved |
Closest papers by meaning
- MARS: A Second-Order Reduction Algorithm for High-Dimensional Sparse Precision Matrices Estimation (JMLR 2023 · similarity 0.85)
- Guarantees of a Preconditioned Subgradient Algorithm for Overparameterized Asymmetric Low-rank Matrix Recovery (ICML 2025 · similarity 0.83)
- Inexact Newton-type Methods for Optimisation with Nonnegativity Constraints (ICML 2024 · similarity 0.83)
- Globally Q-linear Gauss-Newton Method for Overparameterized Non-convex Matrix Sensing (NeurIPS 2024 · similarity 0.83)
- Non-Asymptotic Analysis Of Data Augmentation For Precision Matrix Estimation (NeurIPS 2025 · similarity 0.83)
- RGNMR: A Gauss-Newton method for robust matrix completion with theoretical guarantees (NeurIPS 2025 · similarity 0.82)
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.
