Research map / Optimization / Nonconvex and smooth optimization
A Convergent and Dimension-Independent Min-Max Optimization Algorithm
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 proposed accept-reject min-max algorithm instead of GDA (k=6) | GAN training · Gaussian mixture | fraction of runs generating 4 modes: 0.20 -> 0.70 | Helped |
| Use the proposed accept-reject min-max algorithm instead of GDA (k=1) | GAN training · Gaussian mixture | fraction of runs generating 4 modes: 0 -> 0.70 | Helped |
| Use the proposed accept-reject min-max algorithm instead of OMD | GAN training · Gaussian mixture | fraction of runs generating 4 modes: 0 -> 0.70 | Helped |
| Use the proposed accept-reject min-max algorithm instead of Unrolled-GAN | GAN training · Gaussian mixture | fraction of runs generating 4 modes: 0 -> 0.70 | Helped |
| Use the proposed algorithm instead of OMD | GAN training · CIFAR-10, 50000 iterations | Inception score: 1.96 (0.26) -> 4.68 (0.39) | Helped |
| Use the proposed algorithm instead of GDA | GAN training · CIFAR-10, 50000 iterations | Inception score: 4.51 (0.86) -> 4.68 (0.39) | No clear effect |
| Use the proposed algorithm instead of GDA | GAN training · 01-MNIST | mode collapse frequency | Helped |
| Use the proposed algorithm instead of GDA, OMD, or extra-gradient | Min-max optimization · F1 and F3 test functions | convergence to the global min-max point | Helped |
| Use the proposed algorithm instead of GDA, OMD, or extra-gradient | Min-max optimization · F2 test function | convergence | Hurt |
| Replace the scheduled temperature with a fixed temperature (ablation) | GAN implementation of the proposed algorithm · GAN training | performance | No clear effect |
| Replace randomized acceptance with the deterministic acceptance rule (ablation) | GAN implementation of the proposed algorithm · GAN training | performance | No clear effect |
| Use a fixed number of max-player updates instead of running to a small gradient (ablation) | GAN implementation of the proposed algorithm · GAN training | performance | No clear effect |
Closest papers by meaning
- Minimax Optimization with Smooth Algorithmic Adversaries (ICLR 2022 · similarity 0.91)
- Many Paths to Equilibrium: GANs Do Not Need to Decrease a Divergence At Every Step (ICLR 2018 · similarity 0.89)
- Towards Better Understanding of Adaptive Gradient Algorithms in Generative Adversarial Nets (ICLR 2020 · similarity 0.89)
- Stability and Generalization of Stochastic Gradient Methods for Minimax Problems (ICML 2021 · similarity 0.89)
- GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium (NeurIPS 2017 · similarity 0.88)
- Solving Neural Min-Max Games: The Role of Architecture, Initialization & Dynamics (NeurIPS 2025 · similarity 0.88)
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.