Research map / Learning theory and algorithms / Sample complexity
Distributed Inference for Linear Support Vector Machine
Xiaozhou Wang, Zhuoyi Yang, Xi Chen, Weidong Liu
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 |
|---|---|---|---|
| increase number of aggregation rounds q from 2 to 10 (ablation) | MDL · Simulated SVM n=1e6, p=50, m=1000 | L2 error (x1e-2): 3.223 -> 0.293 | Improved |
| increase number of aggregation rounds q from 2 to 10 (ablation) | MDL · Simulated SVM n=1e6, p=500, m=6250 | L2 error (x1e-2): 3.502 -> 0.320 | Improved |
| add more rounds beyond q=4 (ablation) | MDL · Simulated SVM n=1e6, p=100, m=2000 | L2 error (x1e-2): 0.372 -> 0.338 | No clear effect |
| use MDL multi-round refinement instead of naive divide-and-conquer averaging | MDL · Simulated SVM n=1e4, p=4, m=100 | bias^2 (x1e-4): 59.329 -> 0.004 | Improved |
| use MDL instead of Naive-DC at large batch size | MDL · Simulated SVM n=1e5, p=20, m=5000 | bias^2 (x1e-4): 0.041 -> 0.000 | Mixed |
| increase total sample size n for Naive-DC (ablation) | Naive-DC · Simulated SVM m=100, p=4 | bias^2 (x1e-4): 58.377 -> 60.540 | No clear effect |
| increase total sample size n for MDL (ablation) | MDL · Simulated SVM m=100, p=4 | bias^2 (x1e-4): 1.782 -> 0.000 | Improved |
| increase total sample size n for MDL (ablation) | MDL · Simulated SVM m=100, p=4 | variance (x1e-4): 55.412 -> 1.611 | Improved |
| shrink machine capacity m for Naive-DC (ablation) | Naive-DC · Simulated SVM n=1e4, p=4 | bias^2 (x1e-4): 0.304 -> 59.329 | Hurt |
| vary bandwidth constant C0 in {0.5, 1, 2, 5, 10} (ablation) | MDL · Simulated SVM | L2 error | No clear effect |
Closest papers by meaning
- On Inference for the Support Vector Machine (JMLR 2025 · similarity 0.87)
- Solving L1-regularized SVMs and Related Linear Programs: Revisiting the Effectiveness of Column and Constraint Generation (JMLR 2022 · similarity 0.85)
- Doubly Distributed Supervised Learning and Inference with High-Dimensional Correlated Outcomes (JMLR 2020 · similarity 0.84)
- Divide-and-Conquer for Debiased $l_1$-norm Support Vector Machine in Ultra-high Dimensions (JMLR 2018 · similarity 0.84)
- Optimal Learning Rates for Localized SVMs (JMLR 2016 · similarity 0.84)
- Faster Algorithms for Structured Linear and Kernel Support Vector Machines (ICLR 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.
