Research map / Bandits and online learning / RL theory and MDPs
Provable Model-based Nonlinear Bandit and Reinforcement Learning: Shelve Optimism, Embrace Virtual Curvature
Kefan Dong, Jiaqi Yang, Tengyu Ma
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 ViOlin's virtual ascent with online model learning and reward, gradient, and Hessian prediction | Nonlinear bandit with deterministic reward · Nonlinear bandit | local regret: \calO(\sqrt{T\mathfrak{R}_T}/\epsilon^2) | Improved |
| Use ViOlin for RL with online dynamics learning | Model-based RL with deterministic dynamics and reward · Nonlinear RL | local regret: \calO(\sqrt{T\mathfrak{R}_T}/\epsilon^2) | Improved |
| Use optimism-in-face-of-uncertainty with a two-layer neural net hypothesis class | UCB on a linear ground-truth reward · Linear ground-truth reward, two-layer neural net hypothesis | sample complexity: exponential sample complexity | Hurt |
Closest papers by meaning
- Model-based RL as a Minimalist Approach to Horizon-Free and Second-Order Bounds (ICLR 2025 · similarity 0.89)
- Exploration from a Primal-Dual Lens: Value-Incentivized Actor-Critic Methods for Sample-Efficient Online RL (NeurIPS 2025 · similarity 0.88)
- Small steps no more: Global convergence of stochastic gradient bandits for arbitrary learning rates (NeurIPS 2024 · similarity 0.87)
- Logarithmic Regret for Online KL-Regularized Reinforcement Learning (ICML 2025 · similarity 0.87)
- Kernel-Based Function Approximation for Average Reward Reinforcement Learning: An Optimist No-Regret Algorithm (NeurIPS 2024 · similarity 0.87)
- Online Episodic Convex Reinforcement Learning (ICML 2025 · similarity 0.87)
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.
