Research map / Optimization / Optimizers
Understanding Dynamics of Adam in Zero-Sum Games: An ODE Approach
Yi Feng, Weiming Ou, Xiao Wang
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 smaller first-order momentum beta in Adam-DA (ablation) | Adam-DA · quadratic zero-sum test functions (local convergence) | range of step sizes with local convergence | Improved |
| use larger first-order momentum beta in Adam for minimization (ablation) | Adam (minimization) · quadratic test function | range of step sizes with local convergence | Improved |
| tune beta, rho, epsilon or step size h for Adam-DA on bilinear games | Adam-DA · bilinear games x^T A y | convergence | No clear effect |
| use smaller beta in Adam-DA GAN training (ablation) | Adam-DA · GAN training on CIFAR-10 and STL-10, ResNet and CNN | cumulative average gradient L1 norm | Improved |
| use larger second-order momentum rho in Adam-DA GAN training (ablation) | Adam-DA · GAN training on CIFAR-10 and STL-10, ResNet and CNN | cumulative average gradient L1 norm | Improved |
| lower gradient norm via smaller beta and larger rho (ablation) | Adam-DA GAN training · CIFAR-10 and STL-10 | Inception Score | Improved |
Closest papers by meaning
- Towards Understanding Adam Convergence on Highly Degenerate Polynomials (ICML 2026 · similarity 0.89)
- A Comprehensive Framework for Analyzing the Convergence of Adam: Bridging the Gap with SGD (ICML 2025 · similarity 0.88)
- The Hidden Ratio in Adam: Stable Structure, Compression, and Sign Dynamics (NeurIPS 2026 · similarity 0.88)
- In Search of Adam’s Secret Sauce (NeurIPS 2025 · similarity 0.87)
- Adam under Generalized Smoothness with Second-Moment-Type Stochastic Gradients (NeurIPS 2026 · similarity 0.86)
- On the Implicit Bias of Adam (ICML 2024 · 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.
