Research map / Bandits and online learning / Online convex optimization and games
Doubly Optimal No-Regret Learning in Monotone Games
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 |
|---|---|---|---|
| add Halpern anchoring term (1/(t+1))(x_1 - x_t) to optimistic gradient | optimistic gradient in smooth monotone games · smooth monotone games, last-iterate | last-iterate convergence rate to Nash equilibrium: O(1/sqrt(T)) to O(1/T) | Helped |
| add Halpern anchoring term to optimistic gradient | optimistic gradient in adversarial online learning · adversarial online learning with convex losses | regret | No clear effect |
| adaptive step size driven by second-order gradient variation with threshold 4500πD^2L^2 | AOG · monotone games (self-play) | step size behavior | No clear effect |
| apply extra anchored gradient (EAG) in online learning as a single-call-per-round method | EAG · adversarial online learning (two-player zero-sum construction) | regret: linear Omega(T) regret | Hurt |
| AOG last-iterate rate with dynamic regret compared to prior optimistic/EG methods | individual dynamic regret in monotone games · monotone games | individual dynamic regret: O(√T) to O(log T) | Helped |
| AOG vs EG/OG on convex-concave numerical example | AOG · convex-concave min-max, n=100 bilinear-quadratic | last-iterate convergence and individual dynamic regret | Helped |
Closest papers by meaning
- Uncoupled and Convergent Learning in Monotone Games under Bandit Feedback (NeurIPS 2025 · similarity 0.93)
- Optimism Without Regularization: Constant Regret in Zero-Sum Games (NeurIPS 2025 · similarity 0.91)
- From Average-Iterate to Last-Iterate Convergence in Games: A Reduction and Its Applications (NeurIPS 2025 · similarity 0.90)
- Fast Last-Iterate Convergence of Learning in Games Requires Forgetful Algorithms (NeurIPS 2024 · similarity 0.90)
- Accelerated Regularized Learning in Finite N-Person Games (NeurIPS 2024 · similarity 0.90)
- No-regret Learning in Harmonic Games: Extrapolation in the Face of Conflicting Interests (NeurIPS 2024 · similarity 0.90)
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.