Research map / Bandits and online learning / Online convex optimization and games
Online Policy Gradient for Model Free Learning of Linear Quadratic Regulators with $\sqrt$T Regret
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-free policy gradient with randomized policy-space exploration | LQR control · LQR regret minimization | regret: T^{2/3} -> \sqrt{T} | Helped |
| Use policy-space perturbations whose exploration-cost analysis is quadratic in the radius | Online Policy Gradient · LQR regret minimization | regret: \Oof{\rj^2 \mj} | Helped |
| Estimate gradients from incurred costs without using state observations for parameter tuning | Online Policy Gradient · LQR regret minimization | regret: \OtildeOf[0]{\sqrt{T}} | Helped |
Closest papers by meaning
- Learning Linear-Quadratic Regulators Efficiently with only $\sqrtT$ Regret (ICML 2019 · similarity 0.94)
- Online Linear Quadratic Control (ICML 2018 · similarity 0.93)
- Logarithmic Regret for Learning Linear Quadratic Regulators Efficiently (ICML 2020 · similarity 0.92)
- Efficient Optimistic Exploration in Linear-Quadratic Regulators via Lagrangian Relaxation (ICML 2020 · similarity 0.91)
- Naive Exploration is Optimal for Online LQR (ICML 2020 · similarity 0.91)
- Regret Bounds for Episodic Risk-Sensitive Linear Quadratic Regulator (ICLR 2025 · 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.