Research map / Optimization / Nonconvex and smooth optimization
Accelerated Stochastic Gradient-free and Projection-free Methods
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 Acc-ZO-FW instead of FW-Black | Finite-sum optimization · Finite-sum nonconvex optimization | function query complexity: O(dnε^{-4}) -> O(dnε^{-2}); improves the existing result by a factor of O(ε^{-2}) | Helped |
| Add SPIDER/SpiderBoost variance reduction and momentum acceleration | Acc-SZOFW · Finite-sum nonconvex optimization | function query complexity: O(dnε^{-4}) -> O(dn^{\frac{1}{2}}ε^{-2}); improves the existing result by a factor of O(\sqrt{n}ε^{-2}) | Helped |
| Add SPIDER/SpiderBoost variance reduction and momentum acceleration | Acc-SZOFW · Stochastic nonconvex optimization | function query complexity: O(dε^{-4}) -> O(dε^{-3}); improves the existing result by a factor of O(ε^{-1}) | Helped |
| Replace large-batch SPIDER-style variance reduction with STORM (ablation) | Acc-SZOFW · Stochastic nonconvex optimization | function query complexity: O(dε^{-3}) | No clear effect |
| Replace CooGE with UniGE (ablation) | Acc-SZOFW · Stochastic nonconvex optimization | function query complexity: O(dε^{-3}) | No clear effect |
| Replace CooGE with UniGE (ablation) | Acc-SZOFW* · Stochastic nonconvex optimization | function query complexity: O(dε^{-3}) -> O(d^{\frac{3}{2}}ε^{-3}) | Hurt |
| Use UniGE rather than CooGE (ablation) | Acc-SZOFW variants · Universal adversarial perturbation and robust black-box binary classification | function query convergence speed | Helped |
| Use CooGE rather than UniGE (ablation) | Acc-SZOFW variants · Universal adversarial perturbation | attack loss versus function queries | Hurt |
| Use Acc-SZOFW instead of Acc-SZOFW* | Stochastic zeroth-order Frank-Wolfe · Universal adversarial perturbation | attack loss | Helped |
| Add the proposed momentum acceleration to the deterministic zeroth-order Frank-Wolfe method | Acc-ZO-FW · Single adversarial perturbation (MNIST, CIFAR10) | attack loss | Helped |
| Add variance reduction and momentum acceleration | Acc-SZOFW algorithms · Universal adversarial perturbation and robust black-box binary classification | convergence speed per iteration | Helped |
Closest papers by meaning
- Can Stochastic Zeroth-Order Frank-Wolfe Method Converge Faster for Non-Convex Problems? (ICML 2020 · similarity 0.93)
- Variance-Reduced and Projection-Free Stochastic Optimization (ICML 2016 · similarity 0.90)
- Sarah Frank-Wolfe: Methods for Constrained Optimization with Best Rates and Practical Features (ICML 2024 · similarity 0.90)
- Revisiting Frank-Wolfe for Structured Nonconvex Optimization (NeurIPS 2025 · similarity 0.89)
- Stochastic Zeroth-Order Optimization Under Heavy-Tailed Noise (NeurIPS 2026 · similarity 0.87)
- Zeroth-Order Methods for Constrained Nonconvex Nonsmooth Stochastic Optimization (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.