Research map / Learning theory and algorithms / Kernels and regression theory
Sample-Then-Optimize Batch Neural Thompson Sampling
Zhongxiang Dai, Yao Shu, Bryan Kian Hsiang Low, Patrick Jaillet
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 |
|---|---|---|---|
| remove the gradient term <grad f(x;theta0), theta0'> (Deep Ensemble) (ablation) | STO-BNTS · synthetic GP function | simple regret | Hurt |
| use batch B=4 instead of B=1 | STO-BNTS-Linear · synthetic GP function | simple regret | Improved |
| widen NN from m=64 to m=512 | STO-BNTS-Linear · synthetic GP function | simple regret | Improved |
| narrow NN to m=16 | STO-BNTS-Linear · synthetic GP function | simple regret | Hurt |
| narrow NN to m=16 | STO-BNTS · synthetic GP function | simple regret | Hurt |
| use STO-BNTS-Linear instead of STO-BNTS (ablation) | STO-BNTS · AutoML (RF, XGBoost, CNN tuning) | validation error | Hurt |
| use overly shallow NN (L=1, m=512) instead of L=2, m=256 (ablation) | STO-BNTS · AutoML tasks | validation error | Hurt |
| use overly deep NN (L=8, m=64) instead of L=2, m=256 (ablation) | STO-BNTS · AutoML tasks | validation error | Hurt |
| use batch B=4 instead of B=1 | STO-BNTS · AutoML tasks | validation error | Improved |
| use STO-BNTS instead of Neural UCB, Neural TS, GP-UCB, GP-TS | STO-BNTS · AutoML (RF, XGBoost) | validation error | Improved |
| use STO-BNTS-Linear instead of STO-BNTS (ablation) | STO-BNTS · RL tasks (Lunar-Lander, robot pushing, rover) | reward | No clear effect |
| use batch B=4 instead of B=1 | STO-BNTS · RL tasks | reward | Improved |
Closest papers by meaning
- Thompson Sampling in Function Spaces via Neural Operators (NeurIPS 2025 · similarity 0.89)
- A Study of Bayesian Neural Network Surrogates for Bayesian Optimization (ICLR 2024 · similarity 0.89)
- Bayesian Optimization via Continual Variational Last Layer Training (ICLR 2025 · similarity 0.87)
- Optimizing Posterior Samples for Bayesian Optimization via Rootfinding (ICLR 2025 · similarity 0.87)
- Posterior Sampling-Based Bayesian Optimization with Tighter Bayesian Regret Bounds (ICML 2024 · similarity 0.87)
- Bayesian Optimisation with Unknown Hyperparameters: Regret Bounds Logarithmically Closer to Optimal (NeurIPS 2024 · similarity 0.86)
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.
