Research map / Learning theory and algorithms / Sample complexity
Task-Optimal Exploration in Linear Dynamical Systems
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 task-specific experiment design instead of Frobenius-norm system identification | LQR exploration on the structured system · LQR structured system (varying spectral radius) | LQR excess risk scaling: \mathcal{O} \left ( \tfrac{d_{x}^2}{(1-\rho)^{ 2} } \tfrac{\sigma_w^2}{\gamma^2 T} \right ) versus \Theta \left ( \Big ( | Helped |
| Use task-specific experiment design instead of operator-norm identification | LQR exploration on the structured system · LQR structured system (varying spectral radius) | LQR excess risk scaling: \mathcal{O} \left ( \tfrac{d_{x}^2}{(1-\rho)^{ 2} } \tfrac{\sigma_w^2}{\gamma^2 T} \right ) versus \Theta \left ( \Big ( | Helped |
| Use task-specific experiment design instead of task-optimal Gaussian noise | LQR exploration on the structured system · LQR structured system (varying spectral radius) | LQR excess risk scaling: \mathcal{O} \left ( \tfrac{d_{x}^2}{(1-\rho)^{ 2} } \tfrac{\sigma_w^2}{\gamma^2 T} \right ) versus \Theta \left ( \Big ( | Helped |
| Use task-specific experiment design instead of operator-norm or Frobenius-norm identification | LQR exploration with non-isotropic costs · LQR isotropic system with non-isotropic costs | LQR excess risk scaling: \mathcal{O} \left ( \tfrac{1}{(1-\rho)^4 } \tfrac{\sigma_w^2}{\gamma^2 T} \right ) versus \Theta \left ( \tfrac{d_{x} }{ | Helped |
| Use task-specific experiment design instead of task-optimal Gaussian noise | LQR exploration with non-isotropic costs · LQR isotropic system with non-isotropic costs | LQR excess risk scaling: \mathcal{O} \left ( \tfrac{1}{(1-\rho)^4 } \tfrac{\sigma_w^2}{\gamma^2 T} \right ) versus \Theta \left ( \tfrac{d_{x}^2} | Helped |
| Use a low-regret online LQR strategy to gather data and synthesize a controller | Online LQR controller synthesis · LQR pure exploration after online regret-minimizing data collection | LQR excess risk: \Omega_{\star}(1/\sqrt{T}) | Hurt |
Closest papers by meaning
- Task-agnostic Exploration in Reinforcement Learning (NeurIPS 2020 · similarity 0.90)
- Optimal Exploration for Model-Based RL in Nonlinear Systems (NeurIPS 2023 · similarity 0.89)
- Task-Agnostic Exploration via Policy Gradient of a Non-Parametric State Entropy Estimate (AAAI 2021 · similarity 0.89)
- Geometric Active Exploration in Markov Decision Processes: the Benefit of Abstraction (ICML 2024 · similarity 0.88)
- Decoupling Exploration and Policy Optimization: Uncertainty Guided Tree Search for Hard Exploration (NeurIPS 2026 · similarity 0.88)
- Sample Efficient Myopic Exploration Through Multitask Reinforcement Learning with Diverse Tasks (ICLR 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.