Research map / Probabilistic and scientific modeling / Bayesian optimization
Towards Practical Preferential Bayesian Optimization with Skew Gaussian Processes
Shion Takeno, Masahiro Nomura, Masayuki Karasuyama
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 |
|---|---|---|---|
| Replace LinESS with Gibbs sampling for truncated MVN in MCMC skew GP (ablation) | MCMC-based skew GP estimation · Ackley (d=4) preference model fitting | computational time (sec): about 3x faster (1.61 to 0.54 s) | Improved |
| Replace Gaussian approximation (LA/EP) with exact skew GP via MCMC for predictive duel probabilities | Preferential GP predictions · Ackley (d=4) illustrative example | predictive duel probability accuracy | Hurt |
| Use LA (Laplace approximation) for posterior instead of MCMC | Preferential GP posterior · Ackley illustrative example | posterior mode/mean accuracy | Hurt |
| Hallucination believer: condition GP on a sampled hallucinated v_t and use standard AF (EI or UCB) | Preferential BO (HB-EI, HB-UCB) · 12 benchmark functions (8 shown in main text) | regret vs computational time and iterations | Improved |
| Use bivariate EI or MUC acquisition with hallucination believer (ablation) | HB method · Ackley and Hartmann3 (appendix comparison) | regret | Hurt |
| Use mean maximization as second-point AF in HB (ablation) | HB method · Appendix comparison across benchmarks | regret | No clear effect |
| Use Thompson sampling (DuelTS) instead of HB for second input | Preferential BO on high-dimensional functions · Hartmann4 and Hartmann6 | regret | Hurt |
| Use proposed MC estimators (conditioning on v_t) instead of full MC estimator (ablation) | Posterior statistic estimation for skew GP · Ackley, Holder table, Hartmann6 | RMSE of estimators | Improved |
Closest papers by meaning
- Principled Preferential Bayesian Optimization (ICML 2024 · similarity 0.91)
- Posterior Sampling-Based Bayesian Optimization with Tighter Bayesian Regret Bounds (ICML 2024 · similarity 0.87)
- Standard Gaussian Process is All You Need for High-Dimensional Bayesian Optimization (ICLR 2025 · similarity 0.86)
- Understanding High-Dimensional Bayesian Optimization (ICML 2025 · similarity 0.86)
- Random Exploration in Bayesian Optimization: Order-Optimal Regret and Computational Efficiency (ICML 2024 · similarity 0.86)
- Improved Regret Bounds for Gaussian Process Upper Confidence Bound in Bayesian Optimization (NeurIPS 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.
