Research map / Probabilistic and scientific modeling / Bayesian optimization
Offline Contextual Bayesian Optimization
Ian Char, Youngseog Chung, Willie Neiswanger, Kirthevasan Kandasamy, Andrew Oakleigh Nelson, Mark Boyer, Egemen Kolemen, Jeff Schneider
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 |
|---|---|---|---|
| select tasks via multi-task Thompson sampling (MTS) instead of uniform task selection with standard TS/EI/random | offline contextual Bayesian optimization · synthetic independent tasks (Branin + parabaloids, random 4D/6D) | total simple regret | Improved |
| MTS vs MEI and REVI on correlated finite tasks | offline contextual BO · Branin 1-1, Hartmann 2-2, 3-1, 4-2 | total simple regret | Mixed |
| use stationary RBF kernel with task-adaptive MTS instead of uniform-task TS when task difficulty varies (ablation) | MTS · 10 correlated tasks of variable difficulty | total simple regret | Hurt |
| replace stationary kernel with task-varying lengthscale (Gibbs-style non-stationary) kernel (ablation) | MTS · 10 correlated tasks of variable difficulty | total simple regret | Improved |
| use joint GP over state and action instead of independent GPs per state, with MTS (ablation) | MTS · tokamak fusion simulation, 7 plasma states | total regret | Improved |
| MTS task selection instead of uniform task selection with joint GP (ablation) | joint GP Thompson sampling · tokamak fusion simulation | total regret | Improved |
| MTS task selection instead of uniform selection with independent GPs (ablation) | independent-GP Thompson sampling · tokamak fusion simulation | total regret | Hurt |
Closest papers by meaning
- Optimizing the Unknown: Black Box Bayesian Optimization with Energy-Based Model and Reinforcement Learning (NeurIPS 2025 · similarity 0.82)
- In-Context Black-Box Optimization with Unreliable Feedback (NeurIPS 2026 · similarity 0.81)
- ROOT: Rethinking Offline Optimization as Distributional Translation via Probabilistic Bridge (NeurIPS 2025 · similarity 0.80)
- Stopping Bayesian Optimization with Probabilistic Regret Bounds (NeurIPS 2024 · similarity 0.80)
- Principled Preferential Bayesian Optimization (ICML 2024 · similarity 0.80)
- Bayesian Design Principles for Offline-to-Online Reinforcement Learning (ICML 2024 · similarity 0.80)
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.
