Research map / Bandits and online learning / RL theory and MDPs
The best of both worlds: stochastic and adversarial episodic MDPs with unknown transition
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 performance-difference-based loss shifting in FTRL | Known-transition episodic MDPs with bandit feedback · Known-transition episodic MDP | regret | Helped |
| Replace the prior complex regularizer with Tsallis entropy plus a log-barrier term | Known-transition episodic MDPs with bandit feedback · Known-transition episodic MDP | stochastic regret bound: their U has an extra |A| factor in the first term and an extra L factor in the second term | Helped |
| Use epoch-wise empirical transitions and move optimism into negative loss bonuses | Unknown-transition FTRL · Unknown-transition episodic MDP, adversarial losses | regret | Helped |
| Use a data-dependent learning-rate schedule with Shannon entropy | Full-information FTRL · Unknown-transition episodic MDP, stochastic losses | regret | Helped |
| Set the known-transition method's parameter gamma differently (ablation) | Known-transition Tsallis entropy FTRL · Known-transition episodic MDP, stochastic losses | regret bound dependence on L: L^4 to L^3 | Helped |
| Set the known-transition method's parameter gamma differently (ablation) | Known-transition Tsallis entropy FTRL · Known-transition episodic MDP, adversarial losses | regret | Hurt |
Closest papers by meaning
- Simultaneously Learning Stochastic and Adversarial Episodic MDPs with Known Transition (NeurIPS 2020 · similarity 0.94)
- Dynamic Regret of Adversarial MDPs with Unknown Transition and Linear Function Approximation (AAAI 2024 · similarity 0.91)
- Adapting to Stochastic and Adversarial Losses in Episodic MDPs with Aggregate Bandit Feedback (NeurIPS 2025 · similarity 0.91)
- Beating Adversarial Low-Rank MDPs with Unknown Transition and Bandit Feedback (NeurIPS 2024 · similarity 0.91)
- Near-Optimal Dynamic Regret for Adversarial Linear Mixture MDPs (NeurIPS 2024 · similarity 0.90)
- Online Markov Decision Processes Configuration with Continuous Decision Space (AAAI 2024 · similarity 0.88)
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.