Research map / Learning theory and algorithms / Combinatorial and search optimization
Bayesian Optimization with a Finite Budget: An Approximate Dynamic Programming Approach
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 greedy EI with rollout (h=4, gamma=1.0) lookahead | Bayesian optimization · GP-generated functions mean gap | gap G: 0.762 -> 0.842 | Helped |
| Replace greedy EI with rollout (h=4, gamma=1.0) lookahead | Bayesian optimization · GP-generated functions median gap | gap G: 0.777 -> 0.917 | Helped |
| Use the worst rollout configuration (h=3, gamma=0.9) instead of EI | Rollout · GP-generated functions mean gap | gap G: 0.762 -> 0.767 | Helped |
| Use the worst median rollout configuration (h=4, gamma=0.7) instead of EI | Rollout · GP-generated functions median gap | gap G: 0.777 -> 0.806 | Helped |
| Increase rolling horizon h from 2 to 5 (ablation) | Rollout · GP-generated functions mean gap | gap G: 0.818 -> 0.812 | No clear effect |
| Vary discount factor gamma from 0.5 to 1.0 (ablation) | Rollout · GP-generated functions mean gap | gap G: 0.799 -> 0.842 | No clear effect |
| Replace GP-UCB with rollout (h=4, gamma=1.0) | Bayesian optimization · GP-generated functions mean gap | gap G: 0.711 -> 0.842 | Helped |
| Replace PI with rollout (h=4, gamma=1.0) | Bayesian optimization · GP-generated functions mean gap | gap G: 0.698 -> 0.842 | Helped |
| Replace non-greedy GLASSES with rollout R-4-9 | Bayesian optimization · Test functions (Table 2) | gap G: 5 of 6 metrics better | Mixed |
| Replace PES with rollout | Bayesian optimization · Test functions (Table 2) | gap G: 3 of 8 metrics best for rollout, 3 of 8 for PES | Mixed |
Closest papers by meaning
- BayeSQP: Bayesian Optimization through Sequential Quadratic Programming (NeurIPS 2025 · similarity 0.88)
- Cost-aware Bayesian Optimization via the Pandora's Box Gittins Index (NeurIPS 2024 · similarity 0.87)
- This Too Shall Pass: Removing Stale Observations in Dynamic Bayesian Optimization (NeurIPS 2024 · similarity 0.87)
- Relaxing the Additivity Constraints in Decentralized No-Regret High-Dimensional Bayesian Optimization (ICLR 2024 · similarity 0.86)
- MOBO-OSD: Batch Multi-Objective Bayesian Optimization via Orthogonal Search Directions (NeurIPS 2025 · similarity 0.86)
- Trading Off Quality and Uncertainty Through Multi-Objective Optimisation in Batch Bayesian Optimisation (AAAI 2025 · 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.