Research map / Reinforcement learning / MDPs and planning
The Sample Complexity of Teaching by Reinforcement on Q-Learning
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 |
|---|---|---|---|
| Allow the teacher to override the agent's action and generate arbitrary transitions | Q-learning teaching · MDP family, Level 1 teacher | teaching dimension (TDim): $S$ | Helped |
| Remove the teacher's ability to override the agent's action | Q-learning teaching, relative to Level 1 · MDP family, Level 2 teacher | teaching dimension (TDim): $S$ -> $S(A-1)$ | Hurt |
| Restrict next states to MDP-supported transitions | Q-learning teaching, relative to unconstrained transition control · MDP family, Level 3 teacher | teaching dimension (TDim): $ O\left(SAH\left(\frac{1}{1-\epsilon}\right)^D\right)$ | Hurt |
| Require next states to be sampled from the MDP rather than selected from its support | Q-learning teaching, relative to Level 3 · MDP family, Level 4 teacher | teaching dimension (TDim): $O\left(SAH\left(\frac{1}{(1-\epsilon)p_{\min}}\right)^D\right)$ | Hurt |
| Use stochastic ε-greedy behavior with ε > 0 instead of deterministic behavior | Q-learning with Level 3 teaching · MDP family | teaching dimension (TDim): $[?]$ | Hurt |
| Use deterministic behavior (ε = 0) instead of stochastic ε-greedy behavior | Q-learning with Level 3 teaching · MDP family | teaching dimension (TDim): $TDim= \Theta\left(SAH\right)$ | Helped |
Closest papers by meaning
- PAC Reinforcement Learning With an Imperfect Model (AAAI 2018 · similarity 0.88)
- Machine Teaching for Inverse Reinforcement Learning: Algorithms and Applications (AAAI 2019 · similarity 0.88)
- Sample Efficient Reinforcement Learning with Partial Dynamics Knowledge (AAAI 2024 · similarity 0.88)
- The Sample Complexity of Online Reinforcement Learning: A Multi-model Perspective (ICLR 2026 · similarity 0.88)
- Demonstration-Regularized RL (ICLR 2024 · similarity 0.86)
- Unsupervised Training Sequence Design: Efficient and Generalizable Agent Training (AAAI 2024 · similarity 0.86)
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.