Research map / Probabilistic and scientific modeling / Bayesian optimization
Self-Correcting Bayesian Optimization through Bayesian Active 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 |
|---|---|---|---|
| Use SCoreBO (Hellinger) instead of EI for fully Bayesian BO with GP-sample hyperparameters | EI · 8D GP sample | log regret | Helped |
| Use SCoreBO instead of EI when the BoTorch prior is misaligned | EI · 8D GP sample, unimportant lengthscales | suggested lengthscale: 0.9 -> 1.8 | Helped |
| Use SCoreBO instead of EI/JES/MES on synthetic functions | EI, JES, MES · six noisy synthetic test functions | log inference regret | Helped |
| Use SCoreBO on non-stationary Rosenbrock (4D) | SCoreBO · Rosenbrock 4D | log inference regret | Hurt |
| Use SALSA-HR instead of BALD/BQBC/QBMGP for active learning | BALD, BQBC, QBMGP · six AL functions | negative MLL | Helped |
| Use SAL-HR instead of SAL-WS on RMSE | SAL-WS · six AL functions | RMSE | Hurt |
| Use SCoreBO instead of EI with SAASBO priors on noisy embedded tasks | EI · embedded Ackley-4, Hartmann-6, DNA, SVM HPO | final loss | Helped |
| Use SCoreBO instead of EI at the highest noise level with SAAS prior (ablation) | EI · 25D-embedded Ackley and Hartmann | best observed value | Helped |
| Use SCoreBO instead of EI at noiseless or low noise with SAAS prior (ablation) | EI · 25D-embedded Ackley and Hartmann | best observed value | No clear effect |
| Use SCoreBO instead of EI with AddGP kernel on additive tasks | EI · 6D and 10D GP samples (AddGP) | final value | Helped |
| Use SCoreBO on Cosmological Constants task that is not additively decomposable | SCoreBO · Cosmological Constants (AddGP) | final performance | Hurt |
| Use SCoreBO with HEBO warpings instead of baselines | baselines · PD1 deep learning HPO (3 tasks) | final accuracy | Helped |
Closest papers by meaning
- Informed Initialization for Bayesian Optimization and Active Learning (NeurIPS 2025 · similarity 0.90)
- Understanding High-Dimensional Bayesian Optimization (ICML 2025 · similarity 0.89)
- Bayesian Optimisation with Unknown Hyperparameters: Regret Bounds Logarithmically Closer to Optimal (NeurIPS 2024 · similarity 0.88)
- Bayesian Optimization via Continual Variational Last Layer Training (ICLR 2025 · similarity 0.87)
- Scalable Bayesian Optimization via Focalized Sparse Gaussian Processes (NeurIPS 2024 · similarity 0.87)
- A General Framework for User-Guided Bayesian Optimization (ICLR 2024 · similarity 0.87)
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.