Research map / Bandits and online learning / RL theory and MDPs
Online Learning in Unknown Markov 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 |
|---|---|---|---|
| Use a tunable G ≥ 1 and balance the square-root and linear-in-K terms | V-ol in unknown Markov games · Unknown Markov games | regret against minimax value: O(√(GH^5 SAKι) + KH/G) → Õ(K^{2/3}) (plus the stated H and S factors) | Helped |
| Set G = 1, as in V-sp (ablation) | V-ol in unknown Markov games · Unknown Markov games | regret against minimax value: linear in K | Hurt |
| Learn more aggressively by favoring recent samples with α_t = (GH + 1)/(GH + t), and use the corresponding larger η_t | V-ol relative to V-sp · Online Markov games | regret: linear in K → Õ(K^{2/3}) | Helped |
| Compete with the minimax value instead of the best response in hindsight | Online learning in unknown Markov games · Unknown Markov games | regret benchmark: Ω(min{√(2^H K), K}) | Mixed |
| Avoid maintaining a Q-table and update state values instead | V-ol · Unknown Markov games | dependence of regret on opponent action-space size | Helped |
Closest papers by meaning
- Learning in Markov Games with Adaptive Adversaries: Policy Regret, Fundamental Barriers, and Efficient Algorithms (NeurIPS 2024 · similarity 0.92)
- Policy-Regret Minimization in Markov Games with Function Approximation (ICML 2025 · similarity 0.90)
- Online Markov Decision Processes Configuration with Continuous Decision Space (AAAI 2024 · similarity 0.90)
- Optimism Without Regularization: Constant Regret in Zero-Sum Games (NeurIPS 2025 · similarity 0.90)
- Online Learning with Bounded Recall (ICML 2024 · similarity 0.90)
- 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.