Research map / Optimization / Nonconvex and smooth optimization
Optimal Guarantees for Algorithmic Reproducibility and Gradient Complexity in Convex Optimization
Liang Zhang, Junchi YANG, Amin Karbasi, Niao He
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 quadratic regularization and solve the auxiliary problem with AGD | Convex minimization with inexact initialization · Smooth convex minimization, inexact initialization | reproducibility: \cO(\delta^2) -> \cO(\delta^2) | No clear effect |
| Apply the regularized framework with Inexact-AGD | Convex minimization with inexact deterministic gradients · Smooth convex minimization, inexact gradient | reproducibility: \cO(\delta^2/\epsilon^2) -> \cO(\delta^2/\epsilon^{5/2}) | Hurt |
| Add regularization and use EG to solve the auxiliary minimax problem | Convex-concave minimax optimization · Smooth convex-concave minimax, inexact initialization | reproducibility: \cO(\delta^2 e^{\nicefrac{1}{\epsilon}} \wedge (\delta^2 + 1/\epsilon^2)) -> \cO(\delta^2) | Improved |
| Add regularization and use Inexact-EG for the auxiliary minimax problem | Convex-concave minimax optimization · Smooth convex-concave minimax, inexact gradient | reproducibility: \cO(\delta^2 e^{\nicefrac{1}{\epsilon}} \wedge 1/\epsilon^2) -> \cO(\delta^2/\epsilon^2) | Improved |
| Use a two-loop proximal point framework centered at the previous iterate | Convex-concave minimax optimization · Smooth convex-concave minimax, inexact initialization | reproducibility: \cO(\delta^2) | Improved |
| Use a two-loop proximal point framework centered at the previous iterate | Convex-concave minimax optimization · Smooth convex-concave minimax, inexact gradient | reproducibility: \cO(\delta^2/\epsilon^2) | Improved |
| Use SGDA with stochastic gradients | Stochastic convex-concave minimax optimization · Stochastic smooth convex-concave minimax | sample complexity: \cO(1/\epsilon^2) | No clear effect |
| Use SGDA with stochastic gradients | Stochastic convex-concave minimax optimization · Stochastic smooth convex-concave minimax | reproducibility: \cO\roundBr*{\delta^2/(\epsilon^2 T)} | No clear effect |
Closest papers by meaning
- The Sample Complexity of Gradient Descent in Stochastic Convex Optimization (NeurIPS 2024 · similarity 0.89)
- A Study of First-Order Methods with a Deterministic Relative-Error Gradient Oracle (ICML 2024 · similarity 0.88)
- Last-Iterate Convergence of Single-Loop Stochastic Methods for Constrained Convex-Concave Minimax Problems (NeurIPS 2026 · similarity 0.87)
- High-probability complexity bounds for stochastic non-convex minimax optimization (NeurIPS 2024 · similarity 0.87)
- Adaptive and Optimal Second-order Optimistic Methods for Minimax Optimization (NeurIPS 2024 · similarity 0.87)
- Stochastic Optimization with Arbitrary Recurrent Data Sampling (ICML 2024 · similarity 0.87)
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.
