Research map / Optimization / Optimizers
Structured BFGS Method for Optimal Doubly Stochastic Matrix Approximation
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 |
|---|---|---|---|
| Use structured BFGS instead of dualBB | dual problem for doubly stochastic matrix approximation · Six LIBSVM data sets | running time: 23.01(115) vs 5.10(32); 207.31(162) vs 87.86(54) | Helped |
| Use structured BFGS instead of SSNCG1 | dual problem for doubly stochastic matrix approximation · Six LIBSVM data sets | running time: 24.85(13 ) vs 5.10(32); 846.32(22) vs 87.86(54) | Helped |
| Use s-BFGS instead of alternating projections for affinity normalization | doubly stochastic affinity matrix normalization · gisette, σ = 2.0 | running time: 27.88 vs 1.74; speedup 16.02 | Helped |
| Use s-BFGS instead of alternating projections for affinity normalization | doubly stochastic affinity matrix normalization · mushrooms, σ = 2.0 | running time: 145.87 vs 3.96; speedup 36.84 | Helped |
| Use s-BFGS instead of alternating projections for affinity normalization | doubly stochastic affinity matrix normalization · a6a, σ = 2.0 | running time: 314.58 vs 7.32; speedup 42.97 | Helped |
| Use s-BFGS instead of alternating projections for affinity normalization | doubly stochastic affinity matrix normalization · a7a, σ = 2.0 | running time: 950.22 vs 14.90; speedup 63.77 | Helped |
| Use s-BFGS instead of alternating projections for affinity normalization | doubly stochastic affinity matrix normalization · rcv1.binary, σ = 2.0 | running time: 763.79 vs 27.62; speedup 27.65 | Helped |
| Use s-BFGS instead of alternating projections for affinity normalization | doubly stochastic affinity matrix normalization · a8a, σ = 2.0 | running time: 2713.31 vs 30.61; speedup 88.64 | Helped |
| Use the diagonal Hessian structure and curvature updates rather than a full inverse-Hessian approximation | s-BFGS · synthetic data; varying matrix dimension n | running time | Helped |
Closest papers by meaning
- Stochastic Block BFGS: Squeezing More Curvature out of Data (ICML 2016 · similarity 0.91)
- Accelerated Stochastic Matrix Inversion: General Theory and Speeding up BFGS Rules for Faster Second-Order Optimization (NeurIPS 2018 · similarity 0.89)
- Sharpened Quasi-Newton Methods: Faster Superlinear Rate and Larger Local Convergence Neighborhood (ICML 2022 · similarity 0.88)
- Practical Quasi-Newton Methods for Training Deep Neural Networks (NeurIPS 2020 · similarity 0.87)
- Non-asymptotic Global Convergence Analysis of BFGS with the Armijo-Wolfe Line Search (NeurIPS 2024 · similarity 0.86)
- Affine-Invariant Global Non-Asymptotic Convergence Analysis of BFGS under Self-Concordance (NeurIPS 2025 · similarity 0.85)
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.