Research map / Probabilistic and scientific modeling / Variational and generative sampling
Ensemble Sampling
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 |
|---|---|---|---|
| Ensemble sampling (adding ensemble of models, sampled uniformly) instead of epsilon-greedy | single neuron bandit · single neuron bandit, K=100 actions | instant regret: orders of magnitude lower regret with 50 models | Helped |
| Increase ensemble size from 10 to 50 models (ablation) | ensemble sampling on single neuron bandit · single neuron bandit | instant regret | Helped |
| Ensemble sampling instead of dropout-based approximate Thompson sampling | two-layer neural network bandit · two-layer neural network bandit | instant regret | Helped |
| Dropout-based approximation to posterior (instead of epsilon-greedy) | two-layer neural network bandit · two-layer neural network bandit | instant regret | Hurt |
| Annealing epsilon schedule instead of fixed epsilon in epsilon-greedy (ablation) | two-layer neural network bandit · two-layer neural network bandit | instant regret | Helped |
| Increase ensemble size (more models) for Gaussian bandit (ablation) | ensemble sampling on independent Gaussian bandit · Gaussian bandit 50 arms | per-period regret | Helped |
| Ensemble size needed to approximate Thompson sampling grows with number of actions | ensemble sampling on independent Gaussian bandits · Gaussian bandits varying arms, T=2000, eps=0.03 | minimum number of models | No clear effect |
Closest papers by meaning
- An Analysis of Ensemble Sampling (NeurIPS 2022 · similarity 0.91)
- Thompson Sampling via Local Uncertainty (ICML 2020 · similarity 0.88)
- On Approximate Thompson Sampling with Langevin Algorithms (ICML 2020 · similarity 0.88)
- Feel-Good Thompson Sampling for Contextual Bandits: a Markov Chain Monte Carlo Showdown (NeurIPS 2025 · similarity 0.85)
- Scalable Exploration via Ensemble++ (NeurIPS 2025 · similarity 0.85)
- $\mathtt{VITS}$ : Variational Inference Thompson Sampling for contextual bandits (ICML 2024 · 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.