Research map / Probabilistic and scientific modeling / Bayesian optimization
Scalar Posterior Sampling with Applications
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 |
|---|---|---|---|
| replace dynamic episode schedule (TSDE) with deterministic doubling schedule (DS-PSRL) | PSRL · RiverSwim scalar, experiment 1 | number of policy switches: 159 -> 14 | Helped |
| replace TSDE schedule with DS-PSRL schedule when multi-step exploration is needed | PSRL · RiverSwim scalar, experiment 2 | number of policy switches: 315 -> 14 | Helped |
| use DS-PSRL instead of t-mod-1 and TSDE when contradicting optimal policies make frequent switching oscillate | PSRL · RiverSwim scalar, experiment 2 | average reward | Helped |
| use DS-PSRL instead of TSDE when transition probabilities are identical for all states | PSRL · RiverSwim scalar, experiment 1 | average reward | No clear effect |
| switch to DS-PSRL schedule with more parameters (Dirichlet prior) | PSRL · RiverSwim multiple parameters, 15 states 43 params | number of policy switches: 274 -> 14 | Helped |
| switch to DS-PSRL schedule with more parameters (Dirichlet prior) | PSRL · RiverSwim multiple parameters, 20 states 58 params | number of policy switches: 301 -> 14 | Helped |
| increase number of parameters, DS-PSRL versus TSDE | PSRL · RiverSwim multiple parameters | average reward | Helped |
| switch every time step (t-mod-1) versus DS-PSRL and TSDE schedules | PSRL · LQ problem | average cost | No clear effect |
| use DS-PSRL schedule instead of TSDE in LQ problem | PSRL · LQ problem | number of policy switches: 60 -> 10 | Helped |
Closest papers by meaning
- Towards instance-dependent regret optimality in Episodic MDPs with Posterior Sampling (NeurIPS 2026 · similarity 0.90)
- Model-based Reinforcement Learning for Continuous Control with Posterior Sampling (ICML 2021 · similarity 0.88)
- Exploiting Causal Graph Priors with Posterior Sampling for Reinforcement Learning (ICLR 2024 · similarity 0.85)
- Practical Bayesian Algorithm Execution via Posterior Sampling (NeurIPS 2024 · similarity 0.85)
- Bayesian Optimization from Human Feedback: Near-Optimal Regret Bounds (ICML 2025 · similarity 0.85)
- No-Regret Thompson Sampling for Finite-Horizon Markov Decision Processes with Gaussian Processes (NeurIPS 2025 · similarity 0.84)
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.