Research map / Optimization / Nonconvex and smooth optimization
Single Loop Gaussian Homotopy Method for Non-convex Optimization
Hidenori Iwakiri, Yuhang Wang, Shinji Ito, Akiko Takeda
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 |
|---|---|---|---|
| Use single-loop homotopy ZOSLGHr instead of double-loop ZOGradOpt | ZOSLGH · CIFAR-10 per-image attack | success rate: 65% -> 93% | Improved |
| Use ZOSLGHr instead of ZOGradOpt | ZOSLGH · CIFAR-10 per-image attack | avg iterations to first success: 6789 ± 1901 -> 4979 ± 756 | Improved |
| Use ZOSLGHr instead of ZOSGD | ZOSLGH · CIFAR-10 per-image attack | success rate: 88% -> 93% | Improved |
| Use ZOSLGHr instead of ZOSGD | ZOSLGH · CIFAR-10 per-image attack | avg iterations to first success: 835 ± 1238 -> 4979 ± 756 | Hurt |
| Use ZOSLGHr instead of ZOSGD | ZOSLGH · CIFAR-10 per-image attack | avg total loss: 27.70 ± 74.80 -> 14.26 ± 54.61 | Improved |
| Use ZOSLGHr instead of ZOAdaMM | ZOSLGH · CIFAR-10 per-image attack | avg L2 distortion (successful): 0.050 ± 0.055 -> 0.246 ± 0.178 | Hurt |
| Use ZOSLGHd (derivative-based t update) instead of ZOSLGHr | ZOSLGH · CIFAR-10 per-image attack | avg L2 distortion (successful): 0.246 ± 0.178 -> 0.150 ± 0.084 | Improved |
| Use ZOSLGHd instead of ZOSLGHr | ZOSLGH · CIFAR-10 per-image attack | success rate: 93% -> 92% | No clear effect |
| Use ZOSLGHd instead of ZOSLGHr | ZOSLGH · CIFAR-10 per-image attack | avg total loss: 14.26 ± 54.61 -> 16.49 ± 58.69 | No clear effect |
| Use ZOSLGHd instead of ZOSLGHr | ZOSLGH · CIFAR-10 per-image attack | computational time per iteration | Hurt |
| Use ZOAdaMM instead of ZOSGD | ZO baselines · CIFAR-10 per-image attack | success rate: 88% -> 85% | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- ZOAdaMM — ZO-AdaMM: Zeroth-Order Adaptive Momentum Method for Black-Box Optimization (NeurIPS 2019) · better on 2 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Single-Loop Stochastic Algorithms for Difference of Max-Structured Weakly Convex Functions (NeurIPS 2024 · similarity 0.85)
- Global Optimization with a Power-Transformed Objective and Gaussian Smoothing (ICML 2025 · similarity 0.84)
- High-probability complexity bounds for stochastic non-convex minimax optimization (NeurIPS 2024 · similarity 0.84)
- Stochastic Polyak Step-sizes and Momentum: Convergence Guarantees and Practical Performance (ICLR 2025 · similarity 0.84)
- Last-Iterate Convergence of Single-Loop Stochastic Methods for Constrained Convex-Concave Minimax Problems (NeurIPS 2026 · similarity 0.84)
- Revisiting the Last-Iterate Convergence of Stochastic Gradient Methods (ICLR 2024 · similarity 0.83)
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.
