Research map / Bandits and online learning / Contextual bandits and Thompson sampling
PG-TS: Improved Thompson Sampling for Logistic Contextual Bandits
Bianca Dumitrascu, Karen Feng, Barbara Engelhardt
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 Pólya-Gamma augmented posterior sampling instead of Laplace approximation | Logistic contextual bandits · Gaussian simulation | cumulative regret | Improved |
| Use Pólya-Gamma augmented posterior sampling instead of Laplace approximation | Logistic contextual bandits · Gaussian-mixture simulation | arm selection | Improved |
| Use Pólya-Gamma augmented posterior sampling instead of Laplace approximation | Logistic contextual bandits · Forest Cover Type | cumulative regret | Improved |
| Use Pólya-Gamma augmented posterior sampling instead of Laplace approximation | Logistic contextual bandits · Yahoo! Today Module | CTR | Improved |
| Use one Gibbs sampling iteration per round instead of 100 burn-in iterations (ablation) | PG-TS · Gaussian simulation | cumulative regret | No clear effect |
| Increase the number of burn-in iterations (ablation) | PG-TS · Gaussian simulation | cumulative regret | Improved |
| Increase burn-in beyond 100 iterations (ablation) | PG-TS · Gaussian simulation | cumulative regret | No clear effect |
| Use PG-TS instead of PG-TS-stream | Pólya-Gamma Thompson Sampling · Forest Cover Type | cumulative regret | Improved |
| Use PG-TS-stream instead of Laplace-TS | Logistic contextual bandits · Yahoo! Today Module | CTR | Improved |
| Replace the Gaussian parameter-generating distribution with a four-component Gaussian mixture while retaining the algorithms' Gaussian prior assumption (ablation) | PG-TS and Laplace-TS · Gaussian-mixture simulation | arm selection | Mixed |
Closest papers by meaning
- Feel-Good Thompson Sampling for Contextual Bandits: a Markov Chain Monte Carlo Showdown (NeurIPS 2025 · similarity 0.91)
- $\mathtt{VITS}$ : Variational Inference Thompson Sampling for contextual bandits (ICML 2024 · similarity 0.90)
- Variance-Aware Feel-Good Thompson Sampling for Contextual Bandits (NeurIPS 2025 · similarity 0.90)
- The Choice of Noninformative Priors for Thompson Sampling in Multiparameter Bandit Models (AAAI 2024 · similarity 0.90)
- Logistic Bandits with $\tilde{O}(\sqrt{dT})$ Regret without Context Diversity Assumptions (NeurIPS 2026 · similarity 0.89)
- Thompson Sampling for Multi-Objective Linear Contextual Bandit (NeurIPS 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.
