Research map / Learning theory and algorithms / Matrix and tensor methods
Sparse SVM for Sufficient Data Reduction
Shenglong Zhou
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 sparsity-constrained Newton solver NSSVM instead of SNSVM | SVM solver · higg | TIME (s): 61.75 -> 4.78 | Improved |
| use NSSVM instead of LSVM | SVM solver · higg | TIME (s): 2938 -> 4.78 | Improved |
| use NSSVM instead of FSVM at m=10^8 | SVM solver · Synthetic data m=10^8 | TIME (s): 447.1 -> 12.3 | Improved |
| use NSSVM instead of SNSVM (all samples as SVs) | SVM solver · Synthetic data m=10^8 | NSV/m: 1 -> 1.44e-5 | Improved |
| use NSSVM instead of SNSVM | SVM solver · susy | ACC (%): 78.66 -> 78.82 | Improved |
| use NSSVM instead of LSVM on small-scale dataset | SVM solver · a9a (small-scale) | ACC (%): 84.99 -> 84.04 | Hurt |
| use NSSVM instead of LSVM | SVM solver · skin (large real dataset) | ACC (%): 92.90 -> 91.04 | Hurt |
| use NSSVM instead of LSVM | SVM solver · aips (large real dataset) | ACC (%): 87.66 -> 86.54 | Hurt |
| use NSSVM instead of FSVM | SVM solver · sctp (large real dataset) | ACC (%): 78.01 -> 90.50 | Improved |
| add sparsity level tuning (NSSVM) to fixed-sparsity Alg1 (ablation) | NSSVM · Synthetic data (varying m) | ACC | Improved |
| vary initial point alpha^0 (zero vs random) (ablation) | Algorithm 1 · rlc1-rlc3 and synthetic data | ITER | No clear effect |
| increase sparsity level beta beyond 0.4 (ablation) | Algorithm 1 · Synthetic data | ACC | No clear effect |
Closest papers by meaning
- Scaling Up Sparse Support Vector Machines by Simultaneous Feature and Sample Reduction (ICML 2017 · similarity 0.91)
- Improved Learning Rates of a Functional Lasso-type SVM with Sparse Multi-Kernel Representation (NeurIPS 2021 · similarity 0.88)
- Sparse SVM with Hard-Margin Loss: a Newton-Augmented Lagrangian Method in Reduced Dimensions (JMLR 2025 · similarity 0.88)
- Solving L1-regularized SVMs and Related Linear Programs: Revisiting the Effectiveness of Column and Constraint Generation (JMLR 2022 · similarity 0.88)
- Faster Algorithms for Structured Linear and Kernel Support Vector Machines (ICLR 2025 · similarity 0.86)
- Deep Principal Support Vector Machines for Nonlinear Sufficient Dimension Reduction (ICML 2025 · similarity 0.84)
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.
