Research map / Optimization / Nonconvex and smooth optimization
Unified Methods for Exploiting Piecewise Linear Structure in Convex Optimization
Tyler B Johnson, Carlos Guestrin
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 |
|---|---|---|---|
| add working sets (Algorithm 1) to dual coordinate ascent | DCA · Group lasso, insurance claims, m=1600 | relative suboptimality and support set precision vs time | Improved |
| add working sets (Algorithm 1) to dual coordinate ascent | DCA · Group lasso, m=100 | relative suboptimality and support set precision vs time | Improved |
| add piecewise screening (Theorem 3.4) to dual coordinate ascent | DCA · Group lasso, m=100 | convergence time | Improved |
| add piecewise screening (Theorem 3.4) to dual coordinate ascent | DCA · Group lasso, m=1600 | convergence time | No clear effect |
| scale number of groups m from 100 to 400 to 1600 | DCA + piecewise screening · Group lasso | screening effectiveness: m = 100 -> 1600 | Hurt |
| add working sets (Algorithm 1) to dual coordinate ascent | DCA · Linear SVM, Higgs, m=10^4, 10^5, 10^6 | relative suboptimality vs time | Improved |
| add piecewise screening (Theorem 3.4) to dual coordinate ascent | DCA · Linear SVM, Higgs, large m | relative suboptimality vs time | No clear effect |
| increase number of training examples m | DCA + piecewise screening · Linear SVM, Higgs | fraction of examples screened near best C: m = 10^4 -> 10^6 | Hurt |
| add piecewise screening on top of working sets (ablation) | DCA + working sets · Group lasso and linear SVM | convergence time | No clear effect |
| replace gap screening with piecewise screening (Theorem 3.4) | DCA + screening · Group lasso | convergence time | Improved |
Closest papers by meaning
- Faster Convex Optimization: Simulated Annealing with an Efficient Universal Barrier (ICML 2016 · similarity 0.87)
- Conditional gradient methods for stochastically constrained convex minimization (ICML 2020 · similarity 0.87)
- Universal Gradient Methods for Stochastic Convex Optimization (ICML 2024 · similarity 0.86)
- Barrier Algorithms for Constrained Non-Convex Optimization (ICML 2024 · similarity 0.85)
- Faster Accelerated First-order Methods for Convex Optimization with Strongly Convex Function Constraints (NeurIPS 2024 · similarity 0.85)
- A Universal Transfer Theorem for Convex Optimization Algorithms Using Inexact First-order Oracles (ICML 2024 · 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.
