Research map / Optimization / Nonconvex and smooth optimization
NysADMM: faster composite convex optimization via low-rank approximation
Shipu Zhao, Zachary Frangella, Madeleine Udell
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 |
|---|---|---|---|
| Replace CG with Nystrom-preconditioned PCG as ADMM subproblem solver (NysADMM) with sketch size 50 | inexact ADMM for lasso · lasso, STL-10 and other datasets | solution time (s) at eps=1e-1: NysADMM 165 s vs glmnet 278 s on STL-10 | Improved |
| Use NysADMM instead of SAGA/APG for l1 logistic regression | l1-regularized logistic regression · STL-10, CIFAR-10-rf, p53-rf, connect-4-rf, etc. | solution time (s): e.g. p53-rf 528 s vs SAGA 2116 s and APG 1880 s | Improved |
| Use NysADMM instead of LIBSVM SMO for kernel SVM | RBF kernel SVM · STL-10, realsim-rf, connect-4-rf, etc. | solution time (s): e.g. STL-10 208 s vs LIBSVM 11573 s | Improved |
| Use NysADMM at high precision (eps=1e-2) for lasso | lasso · E2006.train, sector | solution time (s): E2006.train 1657 s vs SSNAL 1446 s | Hurt |
| Sketch size s much smaller than effective dimension (fixed s=50) versus theory-prescribed s ~ deff | NysADMM · lasso and other experiments | solution time / condition number | No clear effect |
| Adaptive sketch size doubling (AdaNysADMM) versus fixed-rank NysADMM | NysADMM · theory only | complexity | No clear effect |
Closest papers by meaning
- Discerning the Linear Convergence of ADMM for Structured Convex Optimization through the Lens of Variational Analysis (JMLR 2020 · similarity 0.89)
- ADMM without a Fixed Penalty Parameter: Faster Convergence with New Adaptive Penalization (NeurIPS 2017 · similarity 0.89)
- Accelerated Variance Reduction Stochastic ADMM for Large-Scale Machine Learning (TPAMI 2020 · similarity 0.89)
- Accelerated Variance Reduced Stochastic ADMM (AAAI 2017 · similarity 0.88)
- ADMM for Nonconvex Optimization under Minimal Continuity Assumption (ICLR 2025 · similarity 0.86)
- ADMM for Structured Fractional Minimization (ICLR 2025 · similarity 0.86)
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.
