Research map / Bandits and online learning / Online convex optimization and games
No-regret learning in games with noisy feedback: Faster rates and adaptivity via learning rate separation
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 |
|---|---|---|---|
| Separate the extrapolation and update learning-rate schedules | OG in variationally stable games with additive noise · Variationally stable games, additive noise | regret: OG: \ding{55}; OG+: \sqrt{\run}\log\run | Helped |
| Separate the extrapolation and update learning-rate schedules | OG in variationally stable games with multiplicative noise · Variationally stable games, multiplicative noise | regret: OG: \ding{55}; OG+: constant | Helped |
| Use dual averaging with separated learning rates | OG+ under adversarial bounded feedback · Adversarial, bounded feedback | regret: OptDA+: \sqrt{\run} | Helped |
| Use adaptive learning rates without prior knowledge of the game or noise profile | OptDA+ in variationally stable games · Additive noise | regret: \bigoh(\sqrt{\nRuns}) | Helped |
| Use adaptive learning rates without prior knowledge of the game or noise profile | OptDA+ in variationally stable games · Multiplicative noise | regret: \bigoh(\exp(1/(2\exponent))) | Helped |
| Use adaptive OptDA+ with positive exponent | OptDA+ against adversarial bounded feedback · Adversarial, bounded feedback | regret: \bigoh(\sqrt{\nRuns}) versus \bigoh(\nRuns^{\frac{1}{2}+\exponent}) | Mixed |
| Use larger exponent in the adaptive learning-rate schedule | Adaptive OptDA+ · Multiplicative noise | regret: \bigoh(\exp(1/(2\exponent))) | Helped |
| Use larger exponent in the adaptive learning-rate schedule | Adaptive OptDA+ · Adversarial, bounded feedback | regret: \bigoh(\nRuns^{\frac{1}{2}+\exponent}) | Hurt |
| Use constant learning rates with OG+ or OptDA+ under multiplicative noise | OG+ and OptDA+ · Multiplicative noise | last-iterate convergence | Helped |
| Use vanishing learning rates under additive noise | OG+ · Additive noise | last-iterate convergence | Helped |
Closest papers by meaning
- Optimism Without Regularization: Constant Regret in Zero-Sum Games (NeurIPS 2025 · similarity 0.91)
- Best of Both Worlds: Regret Minimization versus Minimax Play (ICML 2025 · similarity 0.90)
- Learning in Markov Games with Adaptive Adversaries: Policy Regret, Fundamental Barriers, and Efficient Algorithms (NeurIPS 2024 · similarity 0.90)
- Learning Not to Regret (AAAI 2024 · similarity 0.89)
- Uncoupled and Convergent Learning in Monotone Games under Bandit Feedback (NeurIPS 2025 · similarity 0.89)
- No-regret Learning in Harmonic Games: Extrapolation in the Face of Conflicting Interests (NeurIPS 2024 · 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.