Research map / Optimization / Nonconvex and smooth optimization
Exploiting Strong Convexity from Data with Primal-Dual First-Order Algorithms
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 the data-computed strong-convexity parameter in Opt-BPD instead of setting μ=0 | BPD · Synthetic ridge regression | convergence speed | Helped |
| Adapt parameters using the robust heuristic rather than fixing them | BPD · Synthetic ridge regression | convergence speed | Helped |
| Use adaptive parameters instead of non-adaptive parameters | ADF-SPDC versus DF-SPDC · Synthetic ridge regression, λ=10⁻⁴/n | convergence speed | Helped |
| Use non-adaptive accelerated methods rather than SVRG or SAGA | Katyusha, SPDC, and DF-SPDC · Synthetic ridge regression, λ=10⁻⁴/n | convergence speed | Hurt |
| Use accelerated SPDC variants rather than SVRG or SAGA | SPDC variants · Synthetic ridge regression, λ=10⁻²/n | convergence speed | Helped |
| Use the compared methods with relatively strong regularization | SPDC variants, Katyusha, SVRG, and SAGA · Synthetic ridge regression, λ=1/n | convergence speed | No clear effect |
| Use accelerated algorithms rather than non-accelerated algorithms | SPDC variants and Katyusha versus SVRG and SAGA · cpuact ridge regression | convergence speed | No clear effect |
| Use accelerated algorithms when data strong convexity is weak or absent | SPDC variants and Katyusha · rcv1 and synthetic logistic regression | convergence speed | Helped |
| Use dual-free updates instead of computing the logistic-loss dual proximal mapping | Dual-free SPDC · Logistic regression | convergence speed | No clear effect |
Closest papers by meaning
- Doubly Greedy Primal-Dual Coordinate Descent for Sparse Empirical Risk Minimization (ICML 2017 · similarity 0.90)
- Stochastic Primal-Dual Method for Empirical Risk Minimization with O(1) Per-Iteration Complexity (NeurIPS 2018 · similarity 0.89)
- Variance Reduction via Primal-Dual Accelerated Dual Averaging for Nonsmooth Convex Finite-Sums (ICML 2021 · similarity 0.88)
- Faster Accelerated First-order Methods for Convex Optimization with Strongly Convex Function Constraints (NeurIPS 2024 · similarity 0.87)
- Gradient Descent Converges Arbitrarily Fast for Logistic Regression via Large and Adaptive Stepsizes (ICML 2025 · similarity 0.87)
- Stochastic Smoothed Primal-Dual Algorithms for Nonconvex Optimization with Linear Inequality Constraints (ICML 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.