Research map / Probabilistic and scientific modeling / Bayesian optimization
Batched High-dimensional Bayesian Optimization via Structural Kernel Learning
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 |
|---|---|---|---|
| learn the additive decomposition with Gibbs sampling instead of using no partitions | Add-GP-UCB · Synthetic additive functions, including higher-dimensional settings | regret | Helped |
| learn decompositions with Gibbs sampling instead of randomly sampling candidate decompositions | Add-GP-UCB · Synthetic additive functions | regret | Helped |
| increase the number of observations used for decomposition learning | Gibbs decomposition learner · Synthetic additive functions | decomposition recovery: 0.21 ± 0.13 -> 0.93 ± 0.15 | Helped |
| limit learned group sizes to at most 2 (ablation) | Gibbs decomposition learner on Walker · Walker, 25 parameters | simple regret | Helped |
| use an unconstrained additive decomposition on a function without additive structure (ablation) | Gibbs decomposition learner on Walker · Walker, 25 parameters | simple regret | Hurt |
| use diverse batch selection (PE or DPP) rather than uniform random batch sampling | Batch-UCB methods with known additive structure · Synthetic functions, D=2, 10, 20, and 50 | regret | Helped |
| combine DPP samples by maximizing UCB quality rather than random combination (ablation) | Batch-UCB-DPP · Walker | simple regret | Helped |
| vary the Dirichlet prior scale α (ablation) | Gibbs decomposition learner · Decomposition recovery | Rand Index | No clear effect |
Closest papers by meaning
- Posterior Inference with Diffusion Models for High-dimensional Black-box Optimization (ICML 2025 · similarity 0.90)
- Vanilla Bayesian Optimization Performs Great in High Dimensions (ICML 2024 · similarity 0.90)
- Understanding High-Dimensional Bayesian Optimization (ICML 2025 · similarity 0.90)
- BOIDS: High-Dimensional Bayesian Optimization via Incumbent-Guided Direction Lines and Subspace Embeddings (AAAI 2025 · similarity 0.90)
- Batched Energy-Entropy acquisition for Bayesian Optimization (NeurIPS 2024 · similarity 0.90)
- Automated Kernel Discovery Towards Understanding High-dimensional Bayesian Optimization (NeurIPS 2026 · 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.