Research map / Optimization / Nonconvex and smooth optimization
Competitive Gradient Descent
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 CGD (Nash equilibrium of regularized bilinear local game) instead of GDA/SGA/ConOpt/OGDA | GAN on 2-mode Gaussian mixture · bimodal GAN, eta in {0.4,0.1,0.025,0.005}, RMSProp | training stability (divergence) | Helped |
| use CGD with larger stepsizes instead of OGDA/SGA/ConOpt | covariance estimation (noiseless) · d in {20,40,60} | residual decay per forward evaluation | Helped |
| use CGD at same stepsize as OGDA | covariance estimation (noiseless) · d in {20,40,60} | convergence rate | No clear effect |
| use CGD in stochastic setting | covariance estimation (noisy, d=20) · N in {100,1000,10000} samples | residual | Mixed |
| add competitive term (bilinear interaction) to GDA (ablation) | bilinear game f=alpha*x*y · alpha in {1,3,6}, eta=0.2 | convergence | Helped |
| add consensus term (ConOpt/OGDA style) (ablation) | GDA/LCGD · convex-concave and concave-convex test problems | convergence | Mixed |
| add equilibrium term (matrix inverse) to LCGD, giving CGD (ablation) | LCGD · bilinear game, alpha=6.0 | convergence under strong interaction | Helped |
Closest papers by meaning
- Towards convergence to Nash equilibria in two-team zero-sum games (ICLR 2023 · similarity 0.93)
- Regularized Gradient Descent Ascent for Two-Player Zero-Sum Markov Games (NeurIPS 2022 · similarity 0.92)
- On the $O(1/T)$ Convergence of Alternating Gradient Descent–Ascent in Bilinear Games (ICLR 2026 · similarity 0.92)
- Optimistic Policy Gradient in Multi-Player Markov Games with a Single Controller: Convergence beyond the Minty Property (AAAI 2024 · similarity 0.91)
- Approximating Nash Equilibria in Normal-Form Games via Stochastic Optimization (ICLR 2024 · similarity 0.90)
- Solving Zero-Sum Convex Markov Games (ICML 2025 · similarity 0.89)
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.