Research map / Reinforcement learning / MDPs and planning
Sample-Efficient Reinforcement Learning for Linearly-Parameterized MDPs with a Generative Model
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 model-based planning from anchor-pair transition samples under the linear transition model | Model-based RL · Linear-transition MDP, generative model | sample complexity: \widetilde{O}\left(\frac{K}{\left(1-\gamma\right)^{3}\varepsilon^{2}}\right) | Helped |
| Use vanilla Q-learning with the paper's learning-rate condition | Vanilla Q-learning · Linear-transition MDP, generative model | sample complexity for an ε-accurate Q-function: \widetilde{O}\left(\frac{K}{\left(1-\gamma\right)^{4}\varepsilon^{2}}\right) | Helped |
| Use vanilla Q-learning instead of Phased Parametric Q-learning | Q-learning under the linear transition model · Linear-transition MDP, generative model | sample complexity for an ε-optimal policy: \widetilde{O}\left(\frac{K}{\left(1-\gamma\right)^{6}\varepsilon^{2}}\right) -> \widetilde{O}\left(\frac{K}{\left(1-\gam | Helped |
| Use Optimal Phased Parametric Q-learning with variance reduction instead of vanilla Q-learning | Q-learning under the linear transition model · Linear-transition MDP, generative model | sample complexity: \widetilde{O}\left(\frac{K}{\left(1-\gamma\right)^{4}\varepsilon^{2}}\right) -> \widetilde{O}\left(\frac{K}{(1-\gamma)^{ | Helped |
| Exploit the linear transition structure instead of treating the MDP as tabular | Model-based RL · Linear-transition MDP versus tabular MDP | sample complexity: \widetilde{O}\left(\frac{K}{\left(1-\gamma\right)^{3}\varepsilon^{2}}\right); \widetilde{\Omega}((1-\gamma)^{-3}\varepsi | Helped |
Closest papers by meaning
- Breaking the Sample Size Barrier in Model-Based Reinforcement Learning with a Generative Model (NeurIPS 2020 · similarity 0.95)
- Sample-Efficient Reinforcement Learning Is Feasible for Linearly Realizable MDPs with Limited Revisiting (NeurIPS 2021 · similarity 0.93)
- Optimal Sample Complexity for Average Reward Markov Decision Processes (ICLR 2024 · similarity 0.91)
- Near-Optimal Sample Complexity for MDPs via Anchoring (ICML 2025 · similarity 0.90)
- Near-Optimal Distributionally Robust Reinforcement Learning with General $L_p$ Norms (NeurIPS 2024 · similarity 0.90)
- Finding good policies in average-reward Markov Decision Processes without prior knowledge (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.